Intel P35: Intel's Mainstream Chipset Grows Up
by Gary Key & Wesley Fink on May 21, 2007 3:45 PM EST- Posted in
- CPUs
Synthetic Graphics Performance
The 3DMark series of benchmarks developed and provided by Futuremark are among the most widely used tools for benchmark reporting and comparisons. Although the benchmarks are very useful for providing apples-to-apples comparisons across a broad array of GPU and CPU configurations they are not a substitute for actual application and gaming benchmarks. In this sense we consider the 3DMark benchmarks to be purely synthetic in nature but still very valuable for providing consistent measurements of performance.
In our 3DMark06 test, all of the boards are grouped together with a 2.5% spread from top to bottom. The 975X board takes last place by a small margin, and we noticed in this benchmark that the R600 and 975X do not play well together as the 975X scores within 64 points of the P5K when using a NVIDIA 8800GTX. The latest beta R600 drivers have cured this problem and we will update our scores with the new 8.38 driver set for the roundup. The additional bandwidth available when using the P5K board at 1333/1066 settings resulted in the best CPU and SM2.0 scores in this benchmark.
In the more memory and CPU sensitive 3DMark01 benchmark we see our ASUS P5K Deluxe board taking top honors but only with the FSB set to 1333 (8x333). The Gigabyte P35-DQ6 has the best stock scores at 1066/1066. Our 975x board is slightly handicapped at the 1066/800 settings although this chipset continues to offer excellent memory performance. The higher latencies on the P5K3 board hurt its stock performance although increasing bandwidth to 1333/1333 makes up for it. The spread from top to bottom is again only 2%, however, so minor differences in performance are not really noticeable in either 3DMark.
General System Performance
The PCMark05 benchmark developed and provided by Futuremark was designed for determining overall system performance for the typical home computing user. This tool provides both system and component level benchmarking results utilizing subsets of real world applications or programs. This benchmark is useful for providing comparative results across a broad array of Graphics, CPU, Hard Disk, and Memory configurations along with multithreading results. In this sense we consider the PCMark benchmark to be both synthetic and real world in nature, and it again provides for consistency in our benchmark results.
The ASUS boards have historically done well in this benchmark due to very strong multitasking performance and the same holds true once again. The stock P5K3 scores continue to show an issue with high latencies as this board scored at the bottom in the general application test. The increase to the 1333FSB provides slightly better results with the main increases coming in the multitasking tests. The spread in overall scores is only 1.5%, although in individual areas the differences between the boards may be more or less pronounced.
Rendering Performance
We are using the Cinebench 9.5 benchmark as it tends to heavily stress the CPU subsystem while performing graphics modeling and rendering. Cinebench 9.5 features two different benchmarks with one test utilizing a single core and the second test showcasing the power of multiple cores in rendering the benchmark image. We utilize the standard multiple core benchmark demo and default settings.
The improved bandwidth and front side bus speeds of the P5K3 board results in the top score, putting it 2.5% ahead of the closest competitor for the time being. The remaining boards are all clustered within 2% of each other, with most of the P35 boards placing ahead of the incumbents.
Facebook
Twitter
RSS
58 Comments
View All Comments
Comdrpopnfresh - Tuesday, May 22, 2007 - link
The power could be attributed to the DDR3. With it not being so mature there may be a lot of signaling going on that isn't necessary. Also- with all the new technologies, these boards simply have more going on on them. With more transistors on a cpu its is expected they will use more power- more connections and circuits on a board would mean the same. Everything is running faster too. The power consumption doesn't make sense given the lack of matching real-world performance enhancements, but as the article makes good sense in pointing out, Bios are a big contributing factor here.TA152H - Tuesday, May 22, 2007 - link
Except they ran the power tests with DDR2 on P35 based machines as well, and they were higher than P965 with the same memory. So, obviously, that isn't the cause in this instance.Gary Key - Tuesday, May 22, 2007 - link
After speaking with the board manufacturers and Intel, our original thoughts (briefings/white paper review) were confirmed that the additional circuitry required on the P35 DDR3 boards and in the MCH result in the increased power consumption on the DDR3 platform compared to the DDR2 platform. This holds true for the P35 DDR2 boards when compared to the DDR2 P965, the additional DDR3 circuity/instruction set is still active even though it is not being used. This is why you will see the DDR2/DDR3 combo boards shortly. However, the BIOS engineers believe that can work a little magic with the SpeedStep and C1E wait states to reduce power consumption, however we are talking just a few watts at best. More on this subject in the roundup, at least we hope we will have more... ;)TA152H - Tuesday, May 22, 2007 - link
Gary,Thanks, it's useful to know. Are they going to shackle the x38 with DDR2 support too?
Just confirms my earlier opinion, they should have gotten rid of DDR2 support. Intel is an interesting company, they can come out with a great product like the Core 2, and then have some monkey decide to include DDR2 and DDR3 on the P35. You never know if they'll have a clue, or not. I guess it's a good thing they make turkeys like this and the P7, otherwise we wouldn't have AMD. Although AMD might be the cause of this.
The monkey that decided to do this probably thought, "Oh, look what we can do that AMD can't". It seems to me they did that with the P7, a technological marvel way beyond AMD's capability to design, thank goodness, and the groundbreaking Itanium. Except neither one worked great. AMD's pragmatism has paid off nicely, and even though they can't realistically support DDR2 and DDR3 on the same motherboard, I don't think they really care. Of course, I'm just guessing, when a company does something this stupid, it's always difficult to understand why they did it. It would have been so simple to just have DDR3 support for the P35, and let the P965 handle the DDR2 crowd. It's perfectly adequate.
Thanks again for the information. It's disappointing, but with Intel you get used to it. They can't do everything right after all, and still be Intel.
strikeback03 - Wednesday, May 23, 2007 - link
There might be a more practical reason, such as lack of production capability for DDR3 or HP and Dell threatening to use VIA chipsets instead of P35 in order to keep using DDR2 and keep their prices competitive. I doubt consumers would like their prices increasing by a few hundred dollars for no noticeable performance improvement. And if they only keep the computer 3 or 4 years they will probably spend less on energy than on that DDR3.Who knows about X38, I'd guess DDR2 support won't disappear until the chipset revision for Nehalem.
TA152H - Wednesday, May 23, 2007 - link
Well, I agree if P35 were the only choice from Intel, this would be the case, but again, would you buy VIA if you could get a P965? I wouldn't. If the P965 were a lousy, and seriously obsolete chipset, yes, sure, you'd have to come out with something that replaced it. But they could have easily validated it for FSB of 1333, and at the point the only thing really new in the P35 would be the DDR3 support. So, why would you need it?I was going to get the P35 rather than the x38 because I figure x38 will be even more of a power hog considering the, to me, useless features it has. I don't plan on getting two high-end video cards, and I don't think I will run anything that requires twice the performance of the current PCI-E, but if they drop the DDR2 support, it might the one to go after. If you ever look at an Athlon 64 CPU, you can see the memory controller is simply enormous, so dropping it on the x38 could be significant. With it being high end, they may decide DDR2 isn't a high end technology so they drop it. I hope so.
JarredWalton - Monday, May 21, 2007 - link
Could be the Vista factor? I dunno what else to think about the power numbers.XcomCheetah - Wednesday, May 23, 2007 - link
Could you do a little testing on it... why so high power numbers..Secondly if i remember correctly the power number difference between 680i and P965 chipsets was greater than 20W.. but in your current tests the difference is pretty small.? So any guess what has caused this positive change.?
Reference
http://www.anandtech.com/cpuchipsets/showdoc.aspx?...">http://www.anandtech.com/cpuchipsets/showdoc.aspx?...
http://www.xbitlabs.com/articles/chipsets/display/...">http://www.xbitlabs.com/articles/chipsets/display/...
current power numbers on Anandtech
http://www.anandtech.com/cpuchipsets/showdoc.aspx?...">http://www.anandtech.com/cpuchipsets/showdoc.aspx?...