HASWELL core I5: Performance Preview

[csbin]

Source: http://translate.google.com.hk/tran...ttp://www.chinadiy.com.cn/html/55/n-8455.html

Source: http://translate.google.com.hk/tran...ttp://www.chinadiy.com.cn/html/58/n-8458.html

 

[csbin]

3dmark 11

haswell i5 2.6Ghz W/3G Boost？ Physics Score : 5764

Ivy Bridge i5 3470: Physics Score: 6957

http://www.3dmark.com/3dm11/5742341

 

[Arkaign]

3dmark 11

haswell i5@3G Physics Score : 5764

Ivy Bridge i5 3470: Physics Score: 6957

http://www.3dmark.com/3dm11/5742341

Oh man, if true, that's horrible. 3470 is 3.2Ghz w/3.6 Boost. Judging by those numbers, Haswell has zero IPC boost. Hopefully Haswell is more than just an efficiency boost, but an actual move forward in performance. I have my doubts though.

 

[SPBHM]

GPU score on the second screen shot is basically the same I used to get with a 9600GT.


are you sure it's running at 3GHz when all the cores are being loaded?

 

[csbin]

GPU score on the second screen shot is basically the same I used to get with a 9600GT.


are you sure it's running at 3GHz when all the cores are being loaded?

I am sorry

haswell i5 2.6Ghz W/3G Boost？

http://translate.google.com.hk/translate?hl=zh-CN&sl=zh-CN&tl=en&u=http%3A%2F%2Fwww.chinadiy.com.cn%2Fhtml%2F22%2Fn-8422.html

 

[Madpacket]

Terrible performance..

 

[Enigmoid]

Seems off somehow. How is it getting 42.7 GB/sec bandwidth?

I don't know about the drivers either.

 

[jpiniero]

Seems off somehow. How is it getting 42.7 GB/sec bandwidth?

I don't know about the drivers either.

The article does mention they were able to set the memory to like 2800 Mhz.

 

[Idontcare]

This will make 2500k/2600k owners happy as they can confidently skip 22nm altogether and just wait it out another year to see what 14nm brings.

 

[UaVaj]

 

[386user]

how does 3dmark performance relate to real games performance...? if at all..

 

[Greenlepricon]

I'm gonna wait for some more benchmarks first. I was thinking of getting another pc with an ivy i5 (cause we all need 5 computers right?), but i thought I would just upgrade straight to Haswell. This is a little more depressing than I was hoping it would be though. I don't know if the optimized code will even make this worth it over current cpus for a little while.

 

[Enigmoid]

The article does mention they were able to set the memory to like 2800 Mhz.

That is still unlikely to give 42.7 GB/sec (this is basically 95+% of theoretical compared to about 70-80% for ivy bridge).

128 bit ddr3?

What is impressive is that the system is using less than the whole tdp under load (82 watts).

 

[Homeles]

I just want to know what power and overclocking look like. I already know performance isn't going to be a huge deal.

 

[ZGR]

That is still unlikely to give 42.7 GB/sec (this is basically 95+% of theoretical compared to about 70-80% for ivy bridge).

128 bit ddr3?

What is impressive is that the system is using less than the whole tdp under load (82 watts).

If this is real, then the system must be underclocked, right?

 

[frozentundra123456]

Toms Hardware showed a 3 to 13 percent increase in performance over Ivy. Not sure I would trust any of these early leaks too much. For sure there will not be a spectacular gain, I think that is about all we can say.

 

[Homeles]

Toms Hardware showed a 3 to 13 percent increase in performance over Ivy. Not sure I would trust any of these early leaks too much. For sure there will not be a spectacular gain, I think that is about all we can say.

The only performance preview that matters to me is the AnandTech one

 

[grimpr]

Absolute performance wasnt the priority in Haswell, but mobile performance per watt, graphics and the new pipeline with AVX/FMA3.

 

[Xpage]

I'd wait till actual benchmarks to make a judgement call, but Haswell was made more for mobile than desktop/powerusers/overclockers

 

[WhoBeDaPlaya]

This will make 2500k/2600k owners happy as they can confidently skip 22nm altogether and just wait it out another year to see what 14nm brings.

If it doesn't OC well and get ~5% average IPC increase, I'll just skip down to Microcenter and build a 3930K setup.

 

[Mars999]

If it doesn't OC well and get ~5% average IPC increase, I'll just skip down to Microcenter and build a 3930K setup.

I am with you on that! That new Asus P9X79 WS-E MB looks sweet!!!! and a 3930k. And if HW-E isn't here until 2014 Nov time frame ugh... that 3930k is going to be around for a long time to come then...

 

[BallaTheFeared]

7k at 3.5~GHz? With what ram? 1600 Cas 9?

Um I think my i5-2500k got 10k @ 5.27GHz /w 2200 Ram.

 

[mikk]

3dmark 11

haswell i5 2.6Ghz W/3G Boost？ Physics Score : 5764

Ivy Bridge i5 3470: Physics Score: 6957

http://www.3dmark.com/3dm11/5742341

Once again you cannot compare iGPU 3dmark results with a dedicated 3dmark result system. Yes even the CPU results are boosted a bit from a dedicated GPU system. I could do a 3 Ghz 3dmark run with my HD4000 i5. Maybe later today.

 

[inf64]

Forget the (non-changed) Physics score in Vantage, even worse is the 3dmark11 GPU score of just 530 (compare to 550 on i3 IB with HD4000). I suppose that in 3mdark11 the GPU might have been at 800Mhz instead of maximum of 1200Mhz, but even at 1200Mhz it would score ~800pts in GPU subtest while (now old) Trinity gets 1360pts and Richland will score even more more(~10-15%) on stock. Drivers can be tuned though so I suppose 3dmark11 will be higher on the retail product, but it won't come close to Trinity. Also performance in games might be different than these tests so iGPU performance is still unknown.

 

[Tsavo]

These pre-previews always seem to use the most useless metrics for CPU performance.