2CPU.com » News » November 2005 » Intel Yonah Performance Preview

I don't know how he does it, but Anand has gotten his hands on a 65nm "Yonah" processor and, of course, he has run some benchmarks.Intel, oh Intel, how uninteresting your processors have been to us for so long now. Where have the days of the Northwood gone? Prescott brought us a minor bump in clock speed, minor increases in performance, and more importantly - major increases in power bills. But if any company can go down the wrong path for five years and still come out on top, it

opus13 misanthrope. Posts: 1574 Joined: 2002-04-05

#38438 Posted on: 12/01/2005 01:37 AM for those that really dont want to read the whole article, the result is clearly summarized in 3 letters: 'meh' apparently the only thing yonah does well is to not burn a hole in the carpet.

Kimpsu Registered User Posts: 515 Joined: 2003-01-09

#38440 Posted on: 12/01/2005 02:33 PM Originally posted by opus13: for those that really dont want to read the whole article, the result is clearly summarized in 3 letters: 'meh' apparently the only thing yonah does well is to not burn a hole in the carpet. Really? After reading the article, I thought it did really well. Performance of a 2GHz model is pretty much on par with a 2GHz X2 (on some the Yonah is faster and the X2 on some), consumes less power and it achieved that even while the X2 has the advantage of on die memory controller. 36 watts less power consumption under full load (consumes less under full load than an X2 idle) is good news for laptops, and since it seems to render as fast as (faster and slower than the X2, again depending on the scene rendered), it seems a very good processor for even renderfarms (less heat means you can stuck more of them in the same space than X2s with same level of cooling). And I would like to have a less consuming CPU on my desktop as well since then I could use more silent cooling (as in; less heat requires less cooling / lower rpms). Anyway, the only thing that bothers me is that is Yonah x64 or not?

i_wolf labhair dom as gaelige Posts: 2034 Joined: 2002-11-19

#38441 Posted on: 12/01/2005 02:41 PM I can't say that I honestly agree with you opus. Keep in mind, like anand said this is pre production stuff and performance is likely to change for the better.... prehaps not significantly but it may well improve nonetheless. Even were it not to, the performance of Yonah is fantastic in most applications. At the very elast its a significant performance increase on the existing Intel dual core x86 stuff. Most of the time it sits either slightly above or slightly below an equivalently clocked Athlon 64 x2. Considering that we will be seeing this type of performance in a laptop with unapproachable power requirements... I think its fantastic. Also consider Intel is getting that level of performance without an ondie memory controller... Actually that is one thing I still do not understand with Intel... are they just dragging their feet on this one (ala AMD64 extensions) for the sake of it... why are there no plans to integrate a memory controller on die ??!? Surely this would also allow them to improve overall power efficiency for laptops etc... i_wolf Hung like a donkey. Go like a horse!

Kimpsu Registered User Posts: 515 Joined: 2003-01-09

#38444 Posted on: 12/01/2005 10:35 PM Originally posted by i_wolf: Actually that is one thing I still do not understand with Intel... are they just dragging their feet on this one (ala AMD64 extensions) for the sake of it... why are there no plans to integrate a memory controller on die ??!? Surely this would also allow them to improve overall power efficiency for laptops etc... i_wolf I read somewhere, don't remember where and even if it is true, that on die memory controllers cause problems when the amount of CPUs in a system increases too high... I also remember it said that that is why Opterons are limited to 8 CPUs on a single machine. But as said, I am not sure if it is true.

rmn oh my, it's huge! Posts: 5894 Joined: 2002-01-26

#38445 Posted on: 12/01/2005 11:22 PM You need to make some changes to the cache coherency protocol, but the alternative is to use a single bus for all CPUs, which is far worse. Most high-end systems adopt an intermediate solution (ex., one memory controller for each 4 CPUs / cores). This is what AMD is doing with the Opteron / X2 (1 memory controller per each dual-core CPU). RMN ~~~

opus13 misanthrope. Posts: 1574 Joined: 2002-04-05

#38446 Posted on: 12/02/2005 12:47 AM Originally posted by i_wolf: I can't say that I honestly agree with you opus. Keep in mind, like anand said this is pre production stuff and performance is likely to change for the better.... The reason I say

vedas Registered User Posts: 32 Joined: 2001-12-12

#38447 Posted on: 12/02/2005 01:25 AM Yonah is a chip set to supercede Dothan, so its intended market is for laptops. If you dont think having power concious dual core laptops is a plus, your nuts. Dont say "maybe", say hell yea! AMD will not be delivering this for a while.... If you looking for your desktop CPU extravagansa benchmarks, wait for Conroe to be release before spring, if I rember correct. You confusion products here....

scheme SMP user Posts: 257 Joined: 2002-01-18

#38448 Posted on: 12/02/2005 03:30 AM Originally posted by i_wolf: I can't say that I honestly agree with you opus. Keep in mind, like anand said this is pre production stuff and performance is likely to change for the better.... prehaps not significantly but it may well improve nonetheless. The chips are coming out in about one month. The performance is exactly what you'll see in the final chips. Intel doesn't have enough time to make any changes in the chips and launch it on time. Making new masks would probably take a month so making any changes to the chip now would mean scrapping the launch and delaying a month or two to get production ramped up and the chips revalidated.

Mortimer Registered User Posts: 8 Joined: 2001-07-15

#38450 Posted on: 12/02/2005 07:49 AM Yonah won't be getting any changes, but final chips should run at faster fsb and BIOS settings may change the speed quite a lot from these preproduction boards.

i_wolf labhair dom as gaelige Posts: 2034 Joined: 2002-11-19

#38452 Posted on: 12/02/2005 06:03 PM Yonah won't be getting any changes, but final chips should run at faster fsb and BIOS settings may change the speed quite a lot from these preproduction boards. Exactly my point... again as anand said... it was tested on entirely pre production stuff. There are many things which can affect performance between now and the time it is launched. With regard to performance in many of the tests that the Yonah 'lost' to the Athlon 64 it was within a marginal performance difference which was quite negligable. Its always misleading to say that one platform 'wins' a certain benchmark without taking into account the performance discrepancy between it and its competitor. In which case, for the most part the Yonah does keep up or at least is in the same ballpark as the X2. Some of its 3d S Max bench's were very impressive, given the performance improvement it has made over Dothan in this area. i_wolf Hung like a donkey. Go like a horse!

moshpit Registered User Posts: 28 Joined: 2001-07-22

#38456 Posted on: 12/03/2005 05:15 AM Opus, I REALLY hate defending Intel (I'm particularly AMD biased), but Intel's been saying outright performance increases are to be less of a focus then performance per watt. Yonah is the first REAL showing of this new philosaphy. And it does EXTREMELY well at reflecting the new direction Intel is trying to go. Overall, for a early sample at the low range of the design's clock speed ramp, I think it did excellently against the TRUE king of x86 multicore, X2 and at a significant power savings. Roughly equal performace at much less wattage is perfect against a year older design when you look at it from the perspective of performance per watt and exactly what Intel is shooting for. They hit thier target square on. God, I HATE saying that in Intel's favor. But I gotta. It's true. Quad Damage: Q6600@3.4Ghz/4096mb DDR2@850mhz/Abit IP35 Pro/EVGA 8800GTS 640mb SC/74Gb Raptor/320gb 7200.10 Dual Damage: A64 X2 Toledo@2.62Ghz/TwinX 2x512 XMS3200LL Pro/Asus A8N-SLI Dlx./BFG 7800GT/2x 74Gb WD Raptors

rmn oh my, it's huge! Posts: 5894 Joined: 2002-01-26

#38457 Posted on: 12/03/2005 05:52 AM It's not a "new philosophy". If they could release faster CPUs, they would. Since they can't, they pretend that's not what they were going for anyway. The Pentium-M has always been a great chip, and if Intel USA didn't suffer from such a bad case of "NIH" (not invented here), they would have dumped the P4 and focused on it a long time ago. RMN ~~~

2CPU.com » News » November 2005 » Intel Yonah Performance Preview