After years of being mobile, I finally decided to get myself a cheap but capable desktop from a custom build site. The system is intended to become a home file server, video editing and casual gaming. The specs follows:
MOTHERBOARD: GigaByte GA-EP45T-DS3R Intel P45 Express CrossFireX Chipset LGA775 FSB1600 DDR3/1600 Mainboard w/GbLAN,USB2.0,IEEE1394a,&7.1Audio
CPU: Intel® Core™ 2 Quad Q8200 @ 2.33GHz 1333FSB 4MB L2 **Overclockable S&S*** Cache 64-bit (Read the reviews and I am already hating this CPU)
MEMORY: 4GB PC12800 DDR3/1600mhz Dual Channel Memory (G.SKILL NQ Series w/Heat Spreader ***Overclockable XXX***)
VIDEO: ATI Radeon HD 4870 PCI-E x16 1GB DDR5 Video Card
OPT DRIVE: SONY DUAL FORMAT 20X DVD±R/±RW + CD-R/RW DRIVE DUAL LAYER
HDD: 1TB SATA-II 3.0Gb/s 32M Cache 7200RPM Hard Drive (I know i should get 2 HDDs each for OS and Data)
Total < $1000
OS: Vista Ultimate (i get free SW from MS)
DISPLAY: Dell 22" 1900x1080p S2209W ($236 from Dell)
The Intel Core i7 platform at the moment is quite expensive and it was one of the main reason why I chose the lamest of all the 45nm Quad core. The CPU / Motherboard will be the next upgrade path I take in a couple of years and choosing one at the bottom of the table should hasten the transition. And since it is the video card that does most of the work these days I had to go with AMD's 4870 - best value for money if you ask me.
Feel free to criticize and show how one can do better without breaking the bank.
12.31.2008
New Desktop
Subscribe to:
Post Comments (Atom)
58 comments:
Core i7 is too expensive? I get mine for free ;)
Cheater!
You see, Orthogonal, you can't be rubbing that kind of stuff in an enthusiasts face. It's, well, it's just not right. (You lucky dog!)
SPARKS
Doc, if you’re a builder/enthusiast, sub $1000 is a piece of cake.
Items already purchased:
Case
Power supply
Optical Drive
Hard drives(s)
Mouse + Keyboard
OS (Win Pro)
Needed to purchase:
Core i7 920 300 -------------$300
Motherboard
GIGABYTE GA-EX58-UD4P ------$259
Memory----------------------------$120
http://www.newegg.com/Product/Product.aspx?Item=N82E16820145223
Graphics
EVGA 896-P3-1267-AR GeForce GTX 260 ----$259
http://www.newegg.com/Product/Product.aspx?Item=N82E16814130400
Total---- $979
This thing will kick serious ass and you can add another card later on when finances permit
SPARKS
Why not wait for Phenom II and a nice mobo to overclock it all the way up to 4GHz on air and pair it up with a 4870X2 video card?
The guys over at Xtremesystems are craving for it and it seems that AMD's 45nm process is doing wonders this time (65nm was a fiasco)!
“The guys over at Xtremesystems are craving for it and it seems that AMD's 45nm process is doing wonders this time (65nm was a fiasco)!”
I get your point. Let’s take it a little further.
According to Fuddie, AMD is going to EOL Pheromone II in 2Q or 3Q 2009!?!
“Phenom II X4 940, the 3GHz version of AMD's Deneb core, and Phenom II X4 920 at 2.8 GHz will have an unusually short life cycle. The CPU will officially launch on January 8th and the last orders for these two will have to be placed in mid Q2 2009”
If the demand is there, there won’t be enough to sell. If they were having good yields, there would be plenty to go around. If it were competitive they’d keep building them, say, ¼ the life cycle of a Q6600? (Which got an extension recently)
I am no marketing genius by any stretch (disgusting business), but holy macaroni, they’ve EOL’d the damned things before they’re LAUNCHED!!
GO FIGURE! I can’t.
Maybe they’ve got DDR3 dialed in, finally? Create a high, artificially contrived demand, perhaps? I don’t know.
Pheromone I is/was a drop dead dog. Now, Pheromone II has a six month life cycle. WTF?
http://www.fudzilla.com/index.php?option=com_content&task=view&id=11173&Itemid=1
I don’t know about those idiots at Xtremesystems, but I’ve got a very exclusive $1500 chip that’s been top dog for a year and an 18 month life cycle! They’re lusting for a $300 chip that’s already EOL’d and can barely keep up with a stock clocked QX9770!
Hey, one mans garbage is another mans treasure.
SPARKS
Keep the costs in mind. With Intel keeping the pricing pressure on AMD, the latter is forced to sell their CPUs at near-bargain rates. The Phenom Black Edition costs only $165. Socket AM2 motherboards can run less than $60.
You can get a pretty decent performing system at a very attractive price if you go Phenom. If I was building a second computer, it would be a great option.
"I don’t know about those idiots at Xtremesystems, but I’ve got a very exclusive $1500 chip that’s been top dog for a year and an 18 month life cycle!..."
Funny you saying this because with Phenom II you can use both platforms (AMD+ and AM3). I never trust what TheInquirer or Fudzilla have to say about the top. It's obvious AMD wants to call peoples attention on their AM3 platform, but the 1st iteration of Phenoms will be here for a long long time.
And about "the idiots" at xtremesystems, these guys know what they do. Remember that this site is heavily biased towards intel and when you see such fanboys getting excited about PII, it does tells yous something.
Oh, this little piece made me laugh: "They’re lusting for a $300 chip that’s already EOL’d and can barely keep up with a stock clocked QX9770!
Hey, one mans garbage is another mans treasure."
If all you do is run super pi or cinebench or whatever toy benchmark, then yes, your overpriced "state of the art junk", sorry, I meant processor (hope I didn't hurt your feelings) is well worth all that hard-earned cash. ;)
Anyhow, is up to roborat to decide what suits him best.I guess he's a lot smarter when it comes to spending money :)
SPARKS said .. I don’t know about those idiots at Xtremesystems, but I’ve got a very exclusive $1500 chip that’s been top dog for a year and an 18 month life cycle! They’re lusting for a $300 chip that’s already EOL’d and can barely keep up with a stock clocked QX9770!
i would rather no calling others as idiots, people have their own preferences and budgets. I browsed thru those threads, some are reasonable AMD supporters, and few AMD fanbois and as well as few neutral/intel supporters.
Anonymous Anonymous said...
And about "the idiots" at xtremesystems, these guys know what they do. Remember that this site is heavily biased towards intel and when you see such fanboys getting excited about PII, it does tells yous something.
Nice generalization attempts! :) Look thru their discussion then you know the AMD/Intel/neutral supports and fanbois distribution on that particular discussion! :)
If all you do is run super pi or cinebench or whatever toy benchmark, then yes, your overpriced "state of the art junk", sorry, I meant processor (hope I didn't hurt your feelings) is well worth all that hard-earned cash. ;)
It always make me laugh whenever I see (AMD) fanbois making claim as such, while not providing what are those 'real' applications that they wanna see. Enconder/decoder are not real apps? games are not real apps? Office apps (excel. rar/7z. etc) are not real apps? 3D apps are not real apps? and the list goes on and there are comparison done with these apps (which make those fanbois not so happy anyway)
i really wonder what are the common apps used by those fanbois. :)
I hear that Prime95 is very popular amongst that crowd. :)
I hear that Prime95 is very popular amongst that crowd. :)
Only if it is run for at least 24 hours and you use temperature as your performance indicator! Also if you are OC'ing you should only benchmark using stock cooling ESPECIALLY if you are using a >$500 chip because only an idiot would invest in CPU cooling! Why do those paid pumper Intel sites use alternative cooling solutions....it's not like enthusiasts who are OC'ing actually use them! :)
Blogger Tonus said...
I hear that Prime95 is very popular amongst that crowd. :)
well, with the intention of stressing the CPU to shows its power consumption when their favorite CPU lose out under other load :). It is good for testing stability but not an indicator of 'normal' CPU under load. Who use wprime in daily computing work? :)
and ever wonder how fast a calculation can be done under one CPU to another?
here you go (and those fanbois won't like it:)):
http://forums.techpowerup.com/showthread.php?p=1129032
1. giorgos th. - Core i7 Extreme 965 5004 MHz - 4.959 sec. - Quickest Core i7
2. Binge - Core i7 920 4400 MHz - 5.834
3. Chicken Patty - Core i7 920 4190 MHz - 6.109
4. Jupiter - Core i7 920 4000 MHz - 6.174
5. msgclb - Core i7 920 4074 MHz - 6.374
6. human_error - Core i7 920 3236 MHz - 8.029
7. sno.lcn - Core 2 Extreme QX9650 5040 MHz - 8.547 - Quickest socket 775
8. thraxed - Core i7 920 2834 MHz - 8.915
9. MetalRacer - Core 2 Quad Q9550 4010 MHz - 9.715
10. sno.lcn - Core 2 Quad Q6600 4604 MHz - 9.844
11. DOM - Xeon X3350 3802 MHz - 10.186
12. fatguy1992 - Core 2 Quad Q6600 4005 MHz - 10.249
13. BATOFF - Core 2 Quad Q6600 3915 MHz - 10.500
14. kyle2020 - Core 2 Quad Q6600 4005 MHz - 10.578
15. CrAsHnBuRnXp - Xeon x3350 3800 MHz - 10.601
16. mlee49 - Core 2 Quad Q9450 3760 MHz - 10.843
17. RadeonX2 - Core 2 Quad Q6600 4000 MHz - 10.813
18. dadi_oh - Core 2 Quad Q6600 3960 MHz - 10.922
19. Meizuman - Phenom 9950 3432 MHz 11.234 - Quickest AMD
20. Chicken Patty - Phenom 9950 3412 MHz - 11.250
21. Binge - Core 2 Quad Q9550 3400 MHz - 11.317
22. wolf - Core 2 Quad Q6600 3600 MHz - 11.317
23. trickson - Core 2 Quad Q6600 3601 MHz - 11.390
24. fullinfusion - Phenom 9850 3318 MHz - 11.545
25. Bytor - Phenom 9850 3240 MHz - 11.817
26. servermonkey - Phenom 9850 3311 MHz - 12.218
27. LeMoose - Phenom 9950 3009 MHz - 12.719
28. xubidoo - Core 2 Quad Q6600 3200 MHz - 12.979
The guys over at Xtremesystems are craving for it and it seems that AMD's 45nm process is doing wonders this time (65nm was a fiasco)!
So I did a quick browse over to Xtremesystems and didn't see anyone "craving" for the Phenom II. I saw what I would call a normal level of interest in a new processor, but that is about all.
I also failed to see a bunch of benchmarks that would inspire such enthusiasm. Without some sort of data to back up you suggestion that Robo should consider building a Phenom II system, it is a meaningless suggestion.
So feel free to provide links to back up your claims.
“all you do is run super pi or cinebench or whatever toy benchmark, then yes, your overpriced "state of the art junk", sorry, I meant processor (hope I didn't hurt your feelings) is well worth all that hard-earned cash. ;)”
Ok, I’ll bite.
It’s not the cash, I’ll just make more. It’s like owning the best. The money is just a tool.
I don’t believe SuperPi is an overall indication of a CPU’s performance (As they do on the other sites, ahem.)
But if I were to do comparisons and since you haven’t had the luxury of owning a “state of the art piece of junk” (a contradiction in terms), let me give you a little clue, junior. MY QX9770 (@ 4.27G, Air) will do SuperPi in 10 seconds more than half the time of a “phenomenal” chip, if that’s important to you. This is comparing it to a chip that hasn’t been released yet.
I have had THE ultimate processor for a YEAR. (And in some published benchmarks I still do, like gaming) More importantly, the daily experience in multitasking and overall performance has been worth every penny. (You need to experience one to fully understand.) Game play, overclocking, drive(s) throughput, graphics performance, have been exemplary, daily.
You see, it’s very hard to explain to an arm chair guy, what it’s like to have the best of anything. Like the fastest car, the hottest women, killer sound systems, a model jet, a full blown wood working shop, the hottest CPU’s, the best of anything is going to cost ya. That’s why life in the fast lane in overpriced. Do you know what the fast lane is?
But then again, you sound like the kind of guy who’ll knock the overpriced stuff (all of it) because you’ve never been there, and probably never will. They call it ‘SOUR GRAPES’
I don’t know what your criteria is, performance per watt, performance per dollar, whatever. I’ll take a CORE i7 920 over any garbage AMD is willing to hype or you are willing to buy into. I know what to do with my money. I buy the best, do you?
I hope I didn’t hurt your feelings.
SPARKS
Intheknow: So I did a quick browse over to Xtremesystems and didn't see anyone "craving" for the Phenom II. I saw what I would call a normal level of interest in a new processor, but that is about all.
No, the only P2 craving I see is a bunch of fanbois on Tom's and of course UAEzone :). I suspect Mr. Anonymous is from the latter - they seem to have developed a cadre of unpaid AMD pimpers to go around and try to breathe some life into AMD's CPU business.
"Toy benchmark" is one of the key phrases used over there as well as "Cur i7" or some such doggy reference.
I would suspect Ah-ben-Stoopid or maybe the new idiot, Zooty Tootie but the spelling and sentence structure is too good for them.
"But if I were to do comparisons and since you haven’t had the luxury of owning a “state of the art piece of junk” (a contradiction in terms), let me give you a little clue, junior. MY QX9770 (@ 4.27G, Air) will do SuperPi in 10 seconds more than half the time of a “phenomenal” chip, if that’s important to you. This is comparing it to a chip that hasn’t been released yet..."
Ohh noo, Super Pi again??? :)
Anyhow, if you want to believe in your intel world, fine with me because I'm not the type of guy who push things on peoples throat. Hope you be happy in your fantasy world for ever and ever, because all I'm doing is giving a different point of view for roborat's purchase decission :D
InTheKnow wrote:
So I did a quick browse over to Xtremesystems and didn't see anyone "craving" for the Phenom II. I saw what I would call a normal level of interest in a new processor, but that is about all.
Sorry if I didn't provide any links, but I'll provide you with the links once I get home (I'm typing this at work, lol).
OK, here are some previews:
NOTICE: Results may vary due to different steppings.,
Preview 1
XtremeSystems 1
XtremeSystems 2
This Greek review in progress using actual retail Phenom IIs has been up for well over a week now and indicates that PhII will roughly match Kentsfield at the same clock.
After waiting so long it's pretty disappointing overall if you ask me, as it basically just caught up with a 2-yr old processor that is now at end-of-life. Might as well get a mid-range Yorkfield for roughly the same price, overclock it, and easily beat out an overclocked PhII plus you get SSE4 and lower power consumption.
The Core i7, of course, is generally in another league except for the gaming benches where the results are a bit inconsistent: sometimes a lot faster, sometimes a bit slower than Yorkfield at the same clock.
At the rumored prices, PhII is really only a no-brainer for a current user of Phenom I. Otherwise Yorkfield still appears to be the better value for the mainstream user, and i7 for the power user.
Otherwise Yorkfield still appears to be the better value for the mainstream user, and i7 for the power user.
I agree... there is a reason why Intel is keeping Core2's around for awhile. In a 1 socket system (desktop), IMC and QPI doesn't buy a whole lot except in certain specific applications.
Similarly people have continuously muddled the inclusion of IMC and HT on K8 with the improvements of the core itself. People CHOSE to attribute the lead K8 had to the unique things like IMC and HT, but in reality in desktop space it was the IPC/core of the K8 that was responsible. (I'm talking desktop here, obviously IMC and HT helped a lot in the server space)
It's kind of funny listening to folks talk about the "marginal" improvements of Core i7, as expectations have become a gain from P4 to Core2 for any new architecture. Core i7 appears to be more like a transitionary product for Intel. It catches them up in server space and scaling, and it setups the architecture for integrated GPU and more cores (or even assymetrical cores, given the work Intel is doing with the auto OC?). With the performance of Core2 all Intel really needed to do was validate the IMC and QPI architecture (and we'll see how good/bad it is when the 4P server products come out). My guess is it will not be perfect, but it will not be the failure AMD fans were expecting and the learning curve for Intel on this will be fast
Those same folks talking about the marginal improvements on Core i7 should look very carefully at the actual improvement from a K8 65nm chip to a K10 65nm chip and ask if that also is marginal. You never really see this comparison - it gets mixed in (conveniently) with other variables like 2 core vs 4 core or 65nm vs 45nm.
And I'm not sure what the anonymous links above are supposed to mean... jacking up Vcore to 1.95V and pumping in LN2 is a bit of an academic exercise and is as meaningful as running superpi. I also would be interested in understanding the long term stability (and life expectancy) of the 4GHz air cool OC's with Vcores in the 1.5V area. It is also not using "stock air cooling", as many AMD folks whine about when review sites use non-stock coolers for the Intel OC evaluations. It is funny that it is suddenly not an issue when OC'ing AMD chips. It is also funny that some sort of prime95 torture test is not required to validate the OC when it is an AMD chip, yet over at UAE Zone, an Intel OC is auto-dismissed if they have not done a 24hr prime95 test.
Blogger Axel said...
This Greek review in progress using actual retail Phenom IIs has been up for well over a week now and indicates that PhII will roughly match Kentsfield at the same clock.
It looks like matching (give and take) mainstream Yorkfield to me. from the reviews so far, its OC capability is also matching. Anyway, there is one REAL problem that I see with AMD upcoming PhII (almost wanna type P II :)), is that they use the same numbering like Intel's Core i7, but their performance is not matching at all, especially the 920 compare to the i920 because they are very close in price.
AMD has been riding on others for Marketing ... Athlon XP on Windows XP, Live on Viiv, now confusing buyer with the same model number. Buyers from existing AM2+ owner might not mind about this, but the buyers that intended to build new system and not researching about it will be confused. I would call this semi-cheating.
“Core i7 appears to be more like a transitionary product for Intel. It catches them up in server space and scaling, and it setups the architecture for integrated GPU and more cores (or even assymetrical cores, given the work Intel is doing with the auto OC?). With the performance of Core2 all Intel really needed to do was validate the IMC and QPI architecture”
I love it when you write in technical prose. You bring an anchored sanity to it all.
The last sentence is precisely why I didn’t take the plunge into a i965 setup…… yet. Given the performance of the top end Yorkie there was no need to. It’s simply that good.
However, if I were gearing up for a NEW system as the Doc is, an i920 with an upgradeable X58 platform would be my solution, especially if I were to limit myself to $1K. That $1000 brings a lot of juice to the table, and it leaves (nearly) everything else in its wake.
More importantly, INTC’s product lineup gives a buyer many options as opposed to an AMD PII system which hasn’t been released and is the best and only competitive solution they have, ah---will have.
Oh yeah, the AMD platform is certainly not future proof, in more ways than one.
SPARKS
pointer
It looks like matching (give and take) mainstream Yorkfield to me.
See PetNorth's post near the bottom of this page. He summarizes Yorkfield vs. PhII from the Greek site's results. The rightmost table is compared at about the same clock frequency. With that suite of tests, Yorkie averages ~7% higher IPC. In actuality it's closer to 10% because the 7% figure includes gaming benchmarks at higher resolutions that were GPU bound.
Since Kentsfield's IPC on non-SSE4 apps typically lags behind Yorkfield by only some 5-7%, this actually means that Phenom II IPC hasn't even quite caught up with Kentsfield.
In other words, Phenom II doesn't have a chance at saving AMD's desktop bacon in 2009. In fact, the gap between the 2008 desktop flagships (Agena vs. Yorkfield) and the 2009 flagships (Deneb vs. Nehalem) has in fact widened considerably and AMD will likely be in a worse competitive position going forward. Asset Smart is probably AMD's only hope for surviving 2009.
Sparks: However, if I were gearing up for a NEW system as the Doc is, an i920 with an upgradeable X58 platform would be my solution, especially if I were to limit myself to $1K. That $1000 brings a lot of juice to the table, and it leaves (nearly) everything else in its wake.
That's exactly the same conclusion I came to - drop in a cheap i920 now, OC a bit and then upgrade to a midtier Westmere in a year. I see more DDR3 available now that'll clock up to 1800 or 2K and is compatible with i7.
I really hope somebody investigates what is going on with the i7 gaming benches - why the huge variation between stellar improvement and no improvement performance over Core2 on certain games and multicard setups. Hopefully future games will be written to take advantage of whatever boost i7 can give them.
Axel: In fact, the gap between the 2008 desktop flagships (Agena vs. Yorkfield) and the 2009 flagships (Deneb vs. Nehalem) has in fact widened considerably and AMD will likely be in a worse competitive position going forward.
There's a comparison on pg. 10 of that same thread showing the 2.8 P2 lagging behind a 2.6 i7 by an average score of 21%. Pretty sad for the AMD fans...
Axel wrote:
This Greek review in progress using actual retail Phenom IIs has been up for well over a week now and indicates that PhII will roughly
Sorry but, where does in the entire preview it says it's a retail part??
Also, it seems that these guys got an old revision because the guys at XS have rev 2 and 3 processors all with different steppings.
Funny how you guys take as gospel this single and fishy preview, but at the same time you all question Nehalem's laggin performance in games blaming it on some weird scenarios that only you guys can come up with. What a bunch of hypocrites. :-)
ROBORAT:
You should keep the fanbois from both bands off your blog, they're ruining it. Sooner or later your blog will become the next Sharikou blogspot!
Anonymous, thanks for the links.
I've seen the OCW link in the past. As I recall, the summary was that PhII was a respectable bang for the buck. but the conclusion was also that it was not a top end chip.
With the chip being positioned as a value chip, it is obviously going to be dependent on pricing to be competitive. So while I think it will be successful for AMD, I don't think it solves their monetary problems. Only top end chips are going to provide a premium in this market and AMD needs premium pricing right now.
And let's remain rational about AMD's pricing scheme. They are not my friend any more than Intel is. They will charge what the market will bear. Witness the $1000+ chips when they had the fastest processor around.
The OC on air seems to be quite good, though the LN2 results don't do anything for me. Very few users will ever even try this and OC on LN2 isn't going to make this processor a success.
To address your point about considering a PhII system, it seems that it could be a good decision depending on what you want. The real question in my mind for both Intel and AMD is future compatibility. I'm pretty sure that the current Mobo's for Nehalem will be compatible with Westmere due out in a year. So the investment will pretty much guarantee you leading edge performance for a couple of years at any rate.
I honestly don't know when to believe AMD will come out with their next processor, or what platform it will be part of. AMD seems to be spinning a new "platform" every 3-4 months. This makes me wonder how long my investment is going to last. And, yes, I have seen their roadmap. They just haven't done much recently to inspire confidence in their roadmap. They need to show me consistency to impress me.
Many people point to the high cost of Intel Mobo's. I think you are going to see a change in this dynamic with Westmere. Intel reduces the chipset from 2 chips with 1 for Nehalem. I believe that AMD also only uses a 1 chip solution. I'm pretty comfortable stating that the chipset is the most expensive part of the mobo. So Intel and AMD will now be even in terms of potential solution costs there. Intel is also transitioning chipset production from 8" to 12" processes, so this will give Intel a cost advantage. (Whether or not the consumer will see the price drop is another issue.) The other driver for the cost of Intel boards is layer count. DDR3 necessitated a change from 4 to 6 layer boards. That probably accounts to $20-$40 of the additional cost. AMD boards will have to add that cost when they go to DDR3.
ROBORAT:
You should keep the fanbois from both bands off your blog, they're ruining it. Sooner or later your blog will become the next Sharikou blogspot!
There really aren't any good options for this. You can run an open blog like this one and get all viewpoints at the cost of wading through absurd posts.
You can do what Scientia did and kill your blog with censorship.
Or you can run AMDzone and kill dissenting opinion. I for one would never even bother posting there, because they don't want to hear what I have to say.
I think this blog offers the best solution. I can just ignore the flame bait and take the value from the good posts.
If all you do is run super pi or cinebench or whatever toy benchmark, then yes, your overpriced "state of the art junk", sorry, I meant processor (hope I didn't hurt your feelings) is well worth all that hard-earned cash. ;)
Let's see, Nehalem convinced cray and at least one major animation studio to end exclusive relationships with AMD. If you believe that these companies are influenced by "toy benchmarks" and cinebench then you are welcome to that opinion. Personally, I think this is a telling point, since it hits AMD where they live.
Funny how you guys take as gospel this single and fishy preview, but at the same time you all question Nehalem's laggin performance in games blaming it on some weird scenarios that only you guys can come up with. What a bunch of hypocrites. :-)
Overgeneralization and flamebait.
Perhaps you meet the criteria for banning as a fanboi as proposed by the anonymous poster above?
I too was toying with a 920 (I have a dual core Core2 6700 right now), but I might just go for a cheaper Core2 quad (which would be a drop in replacement on my P35 board) and see what happens with Core i7. I generally like to go one or 2 steps below the top bin and OC if needed.
I don't mind spending the money on DDR3 as most likely the next system build after i7 would also be able to use DDR3. What is keeping me from pulling the trigger on a 920 is the MOBO cost. If I knew the next gen chip would be supported, then I would be more likely to pull the trigger, but I don't know the Intel chipset support plan (I'm talking after Westmere) going forward yet.
Orthogonal said... Core i7 is too expensive? I get mine for free ;)
you lie! i worked for intel for several years and the only way you get your hands on a free CPU's (inlcuding next gen) is if you can "borrow" one from Chandler ATD. ;)
Sparks said...
Total---- $979
Power supplies above 500w, 1TB hard drives and cooler master cases do cost significantly. i needed to buy everything but under $1K. It was impossible with an i7 system.
thanks for the input gents. system bought and should arrive soon.
you lie! i worked for intel for several years and the only way you get your hands on a free CPU's (inlcuding next gen) is if you can "borrow" one from Chandler ATD. ;)
Certain groups can get 'loaner' chips - these are typically engineering samples. Intel 'loans' them to employees, but apparently never asks for them back, though they theoretically could. Presumably this is for tax purposes as giving free chips to employees would probably be considered compensation and force the employee to pay tax on the value of the chip.
“I really hope somebody investigates what is going on with the i7 gaming benches - why the huge variation between stellar improvement and no improvement performance over Core2 on certain games and multicard setups. Hopefully future games will be written to take advantage of whatever boost i7 can give them.”
Moose, I see we’re on the same page. But GURU, with his characteristically low key poignant analysis, as usual, has summed it up in a very tidy nut.
“With the performance of Core2 all Intel really needed to do was validate the IMC and QPI architecture”
They did this quite well, didn’t they?
I’m quite certain the super guru’s at INTC know precisely what the “gaming” issues are. I’m also sure that issue WILL be addressed in future iterations, no doubt. That’s where my money is going. With a New Egg Preferred account, it wasn’t easy holding back the trigger on i965, the fastest processor on the planet.
QX9770 has turned out to be quite a good chip for my application. In fact, it’s getting great mileage as far as this industries’ time line is concerned.
BTW: To those who care. Sure, I’m a card carrying, stock holding, INTEL ever loving nut; I admit it. Hell, I’ve got a well lit curio with every chip I’ve owned since 8086. (Halogen MR 50W) They all bring such fond memories. My children (sigh).
Its presence is obvious upon entering my office, no need to worship, just a small prayer will do.
OXOX
SPARKS
“Certain groups can get 'loaner' chips - these are typically engineering samples. Intel 'loans' them to employees, but apparently never asks for them back, though they theoretically could. Presumably this is for tax purposes as giving free chips to employees would probably be considered compensation and force the employee to pay tax on the value of the chip.”
Foul! Foul! Dammit!
I thought we weren’t going to talk about FREE CHIPS FROM INTC!
YOU’RE KILLING ME.
SPARKS
InTheKnow wrote:
Let's see, Nehalem convinced cray and at least one major animation studio to end exclusive relationships with AMD. If you believe that these companies are influenced by "toy benchmarks" and cinebench then you are welcome to that opinion. Personally, I think this is a telling point, since it hits AMD where they live.
Seems that misinformation runs thoroughly on this blog.
1)Cray considered using intel's chips because of AMD's delayed Budapest processor, which is a Barcelona variant. Adding insult to injury, the TLB bug freaked out many system builders, OEMs and many HPC clients. I would say that Intel had luck on AMD's mess. ;)
2)Funny you mention DreamWorks on this as a Nehalem win because if you don't know, Intel gave Dreamworks those "free Nehalems" to them plus free optimization of their WHOLE software to run good on intel's offerings. I wouldn't say that dreamworks dumped AMD because of their performance, but because they got a way better deal (free processors + software optimizations). :)
1)Cray considered using intel's chips because of AMD's delayed Budapest processor, which is a Barcelona variant. Adding insult to injury, the TLB bug freaked out many system builders, OEMs and many HPC clients. I would say that Intel had luck on AMD's mess. ;)
Clearly, you must work for Cray as you seem to know exactly how/why the decision was made. It is naive to think that Cray changed the entire platform simply due to a short term delay and a bug which was fixed within ~6 months (or do you suppose AMD kept one of their key customers in the dark on the roadmap to fix the TLB bug and the Budapest delay). Considering Nehalem came out well after the TLB fix and Budapest delay (6 months? 9 months?) - are you suggesting Cray originally thought/feared the TLB would never be fixed or that Budapest would slip indefinitely?
It sounds to me as though you are just excuse mongering unless you have some links to support what seems like (biased) conjecture. I suppose Cray did not bother looking at Nehalem performance and we should defer to your logic that Cray simply based their decision on a short term delay and a bug which was fixed on the next stepping.
http://news.cnet.com/8301-13924_3-9985989-64.html
"When we look at the Intel roadmap, it is more closely aligned with our needs," John Batter, president of production at DreamWorks Animations, said during a conversation with Nanotech: The Circuits blog. "The rendering times have been going up because of the complexity and richness of the images. Then you layer on top of that 3D. Something that's already growing--and doubling it."
Intel had the best technology, Batter said. "You need a lot more horsepower. On Intel's upcoming generation, the number of cores is going to help us satiate the big spike in our needs."
Batter specifically mentioned both Nehalem and Larrabee as a reason for the switch to Intel. He said that Larrabee would be "complementary" to Intel's general-purpose CPUs.
Hmmm... should I believe the AMD excuse monger or the president of production at Dreamworks... it's a tough call, but I think I'll choose to trust the Dreamworks guy! :)
Cray considered using intel's chips because of AMD's delayed Budapest processor, which is a Barcelona variant. Adding insult to injury, the TLB bug freaked out many system builders, OEMs and many HPC clients. I would say that Intel had luck on AMD's mess. ;)
I agree with the poster above. Give us links or look like a poser.
Here is my link.
This deal doesn't go down until 2011 or 2012. Are you really so naive that you think that Cray entered an agreement for 3-4 years out based on the TLB bug?
http://www.newser.com/story/31884/dreamworks-jilts-amd-for-intel.html
"For our artists, the impact is going to be really nothing less than monumental," said the studio’s CEO, who called two upcoming generations of Intel chips “radical game-changers” for computing. AMD has worked hard to attract animators as customers, but technical problems with one chip that caused production delays last year have resulted in financial problems for the company.
Of course that is just the studio CEO - what does he know!?!? :) "radical game changers" - I suppose that is a hidden reference to free chips?
Blogger Axel said...
pointer
It looks like matching (give and take) mainstream Yorkfield to me.
See PetNorth's post near the bottom of this page. He summarizes Yorkfield vs. PhII from the Greek site's results.
quite a nice review there. It would be better if they turn on the power saving method for all the tests (or additional test with CnQ/EIST/Cstate on) which are recommended for general users. From some site also saying CnQ is not stable or causes ~10% performance drop, if that's the case, perhaps just turn on that on the Intel system. I do wonder why AMD CnQ support is not that good though. I just registered to that site and going to provide this input once received activation email.
Blogger Roborat, Ph.D said...
Orthogonal said... Core i7 is too expensive? I get mine for free ;)
you lie! i worked for intel for several years and the only way you get your hands on a free CPU's (inlcuding next gen) is if you can "borrow" one from Chandler ATD. ;)
hahaha, I got one Kentsfield and one Yorkfield-extreme 'free' (loan) :) .. and Sparks, i'll probably getting a free Lynnfield next year :)
Anonymous Anonymous said...
you lie! i worked for intel for several years and the only way you get your hands on a free CPU's (inlcuding next gen) is if you can "borrow" one from Chandler ATD. ;)
Certain groups can get 'loaner' chips - these are typically engineering samples. Intel 'loans' them to employees, but apparently never asks for them back, though they theoretically could.
I have seen a case Intel collected it back, on an employee who resigned after loaning a CPU less than 1 year.
yup, I am Intel employee too:) and some standard disclaimer, whatever i said/say/going to say are purely my POV, not representing Intel blahblahblah :)
looking back, I was lured into the blog sphere by Sharikou :). Did not want to disclose my employment because afraid of whatever statement/points I give, good or bad, those fanbois just turn it down by saying i am intel employee, and avoid unnecessary flaming trying to make me to disclose thing that i should not :)
“I would say that Intel had luck on AMD's mess. ;)”
“I wouldn't say that dreamworks dumped AMD because of their performance, but because they got a way better deal (free processors + software optimizations). :)”
“Seems that misinformation runs thoroughly on this blog.”
Dude, dude, …DUDE! Only one of the above statements has a hint of credibility, the last one, and you’re the reason! Your third statement above, almost, but not quite, hints of the monopolistic banter I’m sure your DYING to spew.
Are you calling INTC’s execution of CORE in 2006, Hi-K and 45nM in 2007, and Nehalem and ATOM in 2008, luck!?! Bro, you need to get a reality check. This is the CPU business, not a Vegas crap shoot, ----- hello. INTC has executed FLAWLESSLY!
Conversely, AMD hasn’t done a f**king thing right since BEFORE the purchase of 5.4B ATI (THE most monumental business blunder of historical proportions, ever); nothing, nada, zip.
Wait; there was one thing they did get right. They screwed up a perfectly viable and competitive company.
They lost untold billions
Went the wrong way on process
Simply ignored a duel core market in lieu of a quad core only market
Perfected the term ‘delays’
Went with big name OEMS they could not supply in volume
SCREWED the channel
Diluted the companies value a number of times
Bet the house on a failed design
They even sold the backyard.
Wait, this didn’t happen, it was free chips from Intel that did it.
And Intel got lucky?
No, AMD was the one who got lucky when INTC fell on their asses with Netburst.
Those days are over. This is gospel.
SPARKS
THE most monumental business blunder of historical proportions, ever
Sorry Sparks, I have to give that title to the sub-prime mortgage lenders. We are only now starting to pay for that blunder and who knows how long we will continue to pay.
Hector, for all his efforts has never been able to wreck more than one company at a time. Those jokers have crashed the whole economy.
ITK, I stand corrected. I'll go with 'second most'.
SPARKS
I think it will be a while before we can really see if the ATI purchase was a good or bad move. Right now it looks like a bad move (in spite of the success of the 48xx line) because AMD's goodwill charges indicate that it feels it overpaid by more than $3 billion. Billion. With a 'B'. That's an astronomical sum to overpay!
And so the question is, can the graphics card division generate enough revenue over time to make up for the more than THREE BILLION DOLLARS they overpaid? I don't know the business well enough to make any bold statements, but I figure it will be a while before we can say for sure that it was a mistake or not.
But at the moment, ouch... three billion dollars. Three... billion... dollars...
"and Sparks, i'll probably getting a free Lynnfield next year :)"
I want free chips %(
SPARKS
ITK: I honestly don't know when to believe AMD will come out with their next processor, or what platform it will be part of. AMD seems to be spinning a new "platform" every 3-4 months. This makes me wonder how long my investment is going to last. And, yes, I have seen their roadmap. They just haven't done much recently to inspire confidence in their roadmap. They need to show me consistency to impress me.
Not that I check AMD's roadmap portfolio every day :), but the last time I did check Bulldozer was dozing until sometime in 2011. In fact, the only "bull" that doesn't doze at AMD is their marketing and roadmap crew :). Intel should have a tick and a tock before then (Sandy Bridge is late 2010 I believe). And maybe another tick if the bull dozes until late 2011.
I've been having a sort of argument on Tom's with some AMD fans who claim AMD's financials haven't affected their R&D budget. They base their views on the relative success of the 48xx video cards and the overclocking potential of AMD's 45nm. Of course I point out that AMD gets a lot of help from IBM on the latter, and I'm thinking the die shrink did the ATI design a lot of good for the former. Anyway, I suspect Deneb will be AMD's last gasp for quite some time if not forever, and it isn't the home run AMD needed to gain marketshare or profitability or whatever Hector would like.
As for Mr. Anonymous, I should note that AMD itself merely claims Deneb to be a die shrink with a bigger L3 cache. Anyway, the NDA lifts this coming week I believe so we shall soon have a plethora of legit reviews (and yes, I include Anandtech and Tom's in that category despite the BS to the contrary from AMDZonerz who didn't even merit a freebie eval system from AMD).
“I think it will be a while before we can really see if the ATI purchase was a good or bad move.”
Hmmm, this is an interesting supposition.
The question begs, where would AMD be today, right now, if they didn’t buy ATI?
First, the good-
What was the deal supposed to bring to the table, in theory?
“AMD and ATI joined forces in a transaction that combines AMD’s technology leadership in microprocessors with ATI’s strengths in graphics, chipsets and consumer electronics. The result is a processing powerhouse: a new competitor, better equipped to drive growth, innovation and choice for its customers in commercial and mobile computing segments and in rapidly-growing consumer electronics segments.”
Commercial Clients
Mobile Computing
Consumer Digital Media
High-Growth Markets
http://www.amd.com/us-en/0,,3715_14197_14198,00.html
The bad-
“In a recent interview, the CEO of ATI's long-time rival called the merger a "gift" for NVIDIA, claiming his company's position as the only stand-alone graphics card producer would ensure success in the future.”
http://www.infopackets.com/news/business/amd/2006/20060727_nvidia_overjoyed_with_amd_purchase_of_ati.htm
I’d don’t like the Son-of-a-bitch, but he was right.
“Exclusive PC World tests show that PCs equipped with Intel's new Core 2 Duo processors, formerly code-named Conroe, set new high marks for desktop performance--they're the fastest we've seen by far.”
http://www.pcworld.com/article/126342/intels_new_core_2_duo_processors_run_blazingly_fast_in_pc_world_tests.html
AMD had NOTHING!
The reality-
“Intel's new quad-core Core i7 processors were only launched in late November, but desktop systems featuring the CPUs--from the likes of Dell, Falcon Northwest, Micro Express, and others--have already made an impact on PC World's Top 10 charts.”
http://www.networkworld.com/reviews/2009/010309-intel-core-i7-desktops-dominate.html
“The next installment of CES is under a month away (crazy, right?), and the leaks around the hinges are already starting to show. The latest dirt on the graphical front is this bit from NVIDIA: a twin-GPU behemoth that'll likely destroy anything else on the market today.”
http://www.engadget.com/2008/12/11/twin-gpu-nvidia-geforce-gtx295-expected-at-ces/
Tonus, considering AMD’s financials, their 35% (or less) deal with the ‘Paula’s’ (copyright, GURU), presently, they failed to capitalize on any of the ‘in theory’ benefits above.
What they did was merge (squander billions), take on two extremely powerful competitors, with separate agendas, at precisely the wrong time, at the height of their game.
Skip the ‘theory’ and concentrate on your most valuable asset, your core product. That, in 2006, was the order of the day for AMD.
I submit to you that AMD would be in a far better position today had the ATI deal gone sour. That’s a damned shame.
The graphic componet cannot, under any circumstances, carry AMD. Additionally, there’s a new graphics player entering the arena and I wouldn’t underestimate their potential.
SPARKS
Sparks: I’m quite certain the super guru’s at INTC know precisely what the “gaming” issues are. I’m also sure that issue WILL be addressed in future iterations, no doubt. That’s where my money is going.
I'm thinking we may see some synergistic effect with Larrabee on i7, similar to some of the results shown on multi-nVidia GPU setups for some games. Anyway, methinks I will be lightening my wallet in a couple months on a 920 system - hopefully nVidia will have a die-shrink GPU out by then, or else it'll be a 4870 X2 for me. I had more issues with ATI drivers than nVidia drivers in the past, and would prefer to stick with nVidia for that reason although I think ATI offered more bang per buck.
Hmmm... should I believe the AMD excuse monger or the president of production at Dreamworks... it's a tough call, but I think I'll choose to trust the Dreamworks guy! :)
This was the subject of a thread over on AMDZone by an acknowledged AMD fan who is in the renderfarm industry it seems. He pointed this out to his fellow Zonerz and got flamed to no end for his troubles, mainly by Abinstupid who was ignorantly trying to use his personal H.264 encoding experience on his P1 to the needs of a render farm.
I think Mr. AMD fanbois Anonymous should go back to his home turf and look up that thread - it's only about 6 months old I believe :). Anyway, the point about Dreamworks switching to i7, and other studios considering it as well, was well settled at the end...
Tonus: think it will be a while before we can really see if the ATI purchase was a good or bad move. Right now it looks like a bad move (in spite of the success of the 48xx line) because AMD's goodwill charges indicate that it feels it overpaid by more than $3 billion. Billion. With a 'B'. That's an astronomical sum to overpay!
Reminds me of the famous astronomer Carl Sagan's "Bill-ee-yuns 'n' bill-ee-yuns!", except he was talking about stars and other stellar things, which clearly excludes AMD :).
I saw where AMD is going to write down the ATI purchase once again, but only $70M this time - chump change for sure! :) I think ATI's high point was last quarter where they actually made $50M in profit. Doesn't even cover the interest on the debt, unfortunately. And I suspect nVidia will be coming out with a more competitive GPU forthwith, hopefully just in time for my next build :). 'Twill be lovely to see a GPU price war at that time, seeing as I'm an unlucky bozo who has to pay for everything :).
Sparks: Tonus, considering AMD’s financials, their 35% (or less) deal with the ‘Paula’s’ (copyright, GURU), presently, they failed to capitalize on any of the ‘in theory’ benefits above.
Man, I have to admit - I think Paula Abdul is sexy! :) OK, flame my shorts for it, but she's cute & has a lotta curves in the right places when she doesn't pack on the pounds too much. And it gives me indigestion to think of her and those hairy UAE camel jockies in the same context :).
Yeah I know the UAE guys probably wear business suits instead of robes & burnooses, but still they probably got pointy beards and 4 wives apiece :)
"nVidia will have a die-shrink GPU out by then, or else it'll be a 4870 X2 for me."
Moose, ya can't go wrong with either the X2, or this:
http://www.engadget.com/2008/12/11/twin-gpu-nvidia-geforce-gtx295-expected-at-ces/
As you said, drivers are an issue, but they're refined almost monthly. Price will be a small factor. I suspect they'll be in the same ballpark, however.
The GTX295 will be my personal choice now (on X48), as two these in SLI (X58) would lend itself to a very powerful future upgrade option.
For some reason, SLI scales better in multiple GPU configurations in most games.
SPARKS
“Reminds me of the famous astronomer Carl Sagan's "Bill-ee-yuns 'n' bill-ee-yuns!", except he was talking about stars and other stellar things, which clearly excludes AMD :).”
LOL, remember that? It was the Brooklyn accent slipping through.
It makes me wonder if there’s an equivalent economics parallel to Drake’s Equation regarding the probability of an AMD success. (Sorry that was cold, even for me)
http://www.activemind.com/Mysterious/Topics/SETI/drake_equation.html
SPARKS
The graphic componet cannot, under any circumstances, carry AMD.
AMD would be in a world of hurt without an in house graphics component. We have crossed over from desktop to laptop/mobile as the dominant form factor. (No, I'm not proclaiming the desktop is dead but it isn't growing much.)
Without the ability to do integrated graphics, AMD can't even pretend to compete in this segment of the market. That is why Turion could never get off the ground and AMD knew it. So they needed an in house graphics capability.
The mistake was in overpaying so badly for the capability. Working from the position of having perfect hindsight, a joint venture would have seemed to be a far better approach than tying up all their available cash in the ATI purchase.
Check out the new clock for clock benches by the folks in Greece, Deneb vs. Nehalem vs. Yorkfield vs. Kentsfield all at 3.7 GHz. Also Deneb vs. Agena at 3.0 GHz.
Bottom line according to this review, in general Deneb slightly underperforms Kentsfield at the same clock. However, Deneb is considerably faster than Agena in quite a few benchmarks, especially games.
Also, Deneb's power consumption isn't as great as many previously thought. By total system power draw measured at the wall, at 3.7 GHz even 65-nm Kentsfield draws less power than 45-nm Deneb while generally outperforming it. Nehalem not so much, but then again it performs a lot more work for the power consumed. Finally, Yorkfield is the power consumption champ by far.
The picture might be different at lower clocks, but at 3.7 GHz it doesn't look that good for Deneb.
Also, regarding Nehalem's impact on render farms and HPC, I think people are in for some big surprises when the server Nehalem NDAs are lifted. We've already seen some sneak peek previews with SPEC, SAP, and HPC benchmarks. They seem to indicate that 2P Nehalem may generally outperform 4P Barcelona and perhaps even be competitive with 4P Shanghai.
Sparks: The GTX295 will be my personal choice now (on X48), as two these in SLI (X58) would lend itself to a very powerful future upgrade option.
For some reason, SLI scales better in multiple GPU configurations in most games.
Whoa, that looks impressive! I wonder how many megawatts PSU I'll need :). Seriously, I hope 55nm does nVidia some good on the power consumption here, as my old GTX8800 pumps a lot of hot air out into my basement "office" which has rather poor airflow. Great at the moment but not during summertime.
As for Drake's equation, IIRC one of the factors was the probability of intelligent life arising on a planet. I'd guess any AMD-equivalent equation would need a probability factor of intelligent life arising on UAEZone :).
Axel: Bottom line according to this review, in general Deneb slightly underperforms Kentsfield at the same clock. However, Deneb is considerably faster than Agena in quite a few benchmarks, especially games.
Yeah I'be been following that on the Xtremesystems thread, http://www.xtremesystems.org/forums/showthread.php?t=212641&page=10, since the Greek-English translator gives me a headache :). Seems a 2.6 i7 920 is on average about 21.6% faster than a 2.8GHz P2, which accounts for a one-speed-bin turbomode boost on the i7 to yield rough clock-to-clock comparisons.
Post a Comment