I7 4770K Benchmarks

[csbin]

http://translate.google.com.hk/tran...um.php?mod=viewthread&tid=12622&extra=&page=1

 

[mikk]

Low Cinebench score. Maybe CPU Turbo didn't work?

 

[Charles Kozierok]

Are these benchmarks trustworthy? Who is doing them? What is their track record?

 

[Ed1]

the 4570 there shows only 3.2ghz verse 3570 is 3.4ghz , not sure whats up with that .
the 4770 seems right (same as 3770)

 

[frozentundra123456]

Are these benchmarks trustworthy? Who is doing them? What is their track record?

Will wait for full set of benchmarks from a reliable site. Who knows what this means. Not that I am expecting anything great, but benchmarks on real shipping silicon should be out in a few weeks.

 

[Ed1]

Will wait for full set of benchmarks from a reliable site. Who knows what this means. Not that I am expecting anything great, but benchmarks on real shipping silicon should be out in a few weeks.

right, would also be good to have baseline 3770/3570 results on same system (as much as possible as MB is different) .

 

[inf64]

Low Cinebench score. Maybe CPU Turbo didn't work?

Turbo is working. Check OCed 4770K @4.5Ghz results, that's fixed OC clock. The difference between OC 4770K and stock one mirrors the difference between 3.8-3.9Ghz and 4.5Ghz . Stock Haswell runs at 3.8-3.9Ghz clock in most of those workloads (even MTed ones).

Results are in line with previous leaks, nothing out of the ordinary. IPC improvement is not worth mentioning unless one uses AVX2 or FMA (in which case it's very noticeable performance increase).

 

[mikk]

Turbo is working. Check OCed 4770K @4.5Ghz results, that's fixed OC clock. The difference between OC 4770K and stock one mirrors the difference between 3.8-3.9Ghz and 4.5Ghz . Stock Haswell runs at 3.8-3.9Ghz clock in most of those workloads (even MTed ones).

I meant his default Benchmarks results and not his fixed 4.5 Ghz.

 

[inf64]

Well read what I wrote. You can actually see that Turbo is working by comparing the 4.5Ghz results and stock ones. The performance delta is equal or very close to the difference between 4.5Ghz and 3.8-3.9Ghz clock (the stock range Haswell runs at in most workloads).

 

[scannall]

While these kinds of posts are fun and all, I really don't attach any real meaning to them. Real benchmarks, from reputable sites will be along soon enough. Then the arguing can begin in earnest.

 

[ShintaiDK]

the 4570 there shows only 3.2ghz verse 3570 is 3.4ghz , not sure whats up with that .
the 4770 seems right (same as 3770)

4670 is the successor to 3570.

 

[mikk]

Well read what I wrote. You can actually see that Turbo is working by comparing the 4.5Ghz results and stock ones. The performance delta is equal or very close to the difference between 4.5Ghz and 3.8-3.9Ghz clock (the stock range Haswell runs at in most workloads).

There is no Cinebench 11.5 4.5 Ghz result. How do you compare?

The only other multithreaded Benchmark result from the 4.5 Ghz Haswell is Fritzchess. The score is 21,3% over i7-4770k default. That doesn't make sense. Multithread Turbo from i7-4770k is 3,9 Ghz, means 4.5 Ghz clocks 15,4% higher. 21,3% higher score while clocked only 15,4% higher (assuming the multithread Turbo worked on Haswell) does not make sense to me. And also I don't believe Haswell is slower in Cinebench and Fritzchess.

 

[Hulk]

0.88V @ 3.9GHz?

 

[Rvenger]

I might just get one of these to check out the new premium motherboards they have. Unless I need AVX2 this seems like a sidegrade to my 3770k. That chip is really scratched above so I wouldn't base CPU temps on that if there were any. lol

 

[blackened23]

0.88V @ 3.9GHz?

Hmm, sounds slightly lower than Ivy Bridge' 1V at idle. Should be interesting to see how power consumption is affected in normal desktop usage.

 

[blackened23]

I might just get one of these to check out the new premium motherboards they have. Unless I need AVX2 this seems like a sidegrade to my 3770k. That chip is really scratched above so I wouldn't base CPU temps on that if there were any. lol

Remind me, does Z87 bring anything new to the table? SATA express perhaps? Of course this is a moot point since a new motherboard is required for Haswell anyway, but i'm just curious.

 

[TidusZ]

this seems like a sidegrade to my 3770k

which was a sidegrade to a 2600k

 

[ShintaiDK]

Remind me, does Z87 bring anything new to the table? SATA express perhaps? Of course this is a moot point since a new motherboard is required for Haswell anyway, but i'm just curious.

6 SATA6, 2 more USB3, fast UEFI boot, SFDP, quad read SPI.

 

[inf64]

There is no Cinebench 11.5 4.5 Ghz result. How do you compare?




The only other multithreaded Benchmark result from the 4.5 Ghz Haswell is Fritzchess. The score is 21,3% over i7-4770k default. That doesn't make sense. Multithread Turbo from i7-4770k is 3,9 Ghz, means 4.5 Ghz clocks 15,4% higher. 21,3% higher score while clocked only 15,4% higher (assuming the multithread Turbo worked on Haswell) does not make sense to me. And also I don't believe Haswell is slower in Cinebench and Fritzchess.

MTed Turbo is not guaranteed to be 3.9GHz. That is maximum Turbo clock.

Stock Vs OC
3dmark11 physics test: 34.84pts Vs 39.23pts ->12.6% perf. increase (15.3% clock increase from 3.9Ghz- in this test Haswell runs at max. Turbo)
SuperPi 1M: 9.3s Vs 8s ( 8/9.3=0.86 or 14% performance increase with 15% clock increase- in this test Haswell runs at max. Turbo)
Fritz: 14.39 vs 17.4 ( 20% performance increase which is a good indicator Haswell @ stock is not able to run at max. Turbo of 3.9Ghz;it's between 3.7 and 3.8Ghz in this benchmark that stresses all cores and is fp/simd intensive)
aida L1 cache read : 124776 MB/s vs 143964 MB/s ( 15.3% performance increase with 15.3% clock increase).

C11.5 is missing but all other data aligns with the clock difference.
Turbo is working properly.

0.88V @ 3.9GHz?

Power management at work. The clock is not 3.9Ghz, it's idle clock (whatever Haswell runs at in idle mode). CPUz is just not picking it up properly, happens all the time.

 

[ShintaiDK]

Be careful what boards you compare with. A lot of Z77 boards will turbo 4 bins instead of 2 as default under full load.

 

[mikk]

MTed Turbo is not guaranteed to be 3.9GHz. That is maximum Turbo clock.

It seems you are not familiar with Intel CPUs. Multithread Turbo is always active on Intel desktop CPUs especially on such short non demanding CPU Benchmarks like Cinebench or Fritzchess. And some Z77 Boards use an even higher frequency @default. 3dmark11 is a bad CPU test, SuperPi 1M is too short. Only Fritzchess and Cinebench are meaningful. The slower Fritzchess result doesn't make sense. The pre-release non-professional testers are really pathetic nowadays. They should disable the turbo and should make sure the CPU runs at base frequency all the time. They can't handle the CPU Turbo or auto overlock features by some Z77 boards properly.

 

[A5]

6 SATA6, 2 more USB3, fast UEFI boot, SFDP, quad read SPI.

Yeah, its not a big step like Z77 was. Quad SPI could be nice I guess - make flashing a bit faster

Rumors have SATA Express slated for a hypothetical Z97 , right?

 

[AdamK47]

Anyone know what the max bandwidth is for the SATA-III controller on Z87?

 

[ShintaiDK]

Yeah, its not a big step like Z77 was. Quad SPI could be nice I guess - make flashing a bit faster

Rumors have SATA Express slated for a hypothetical Z97 , right?

Z97/H97 is not hypethetical. And yes, SATA Express there

 

[inf64]

It seems you are not familiar with Intel CPUs. Multithread Turbo is always active on Intel desktop CPUs especially on such short non demanding CPU Benchmarks like Cinebench or Fritzchess. And some Z77 Boards use an even higher frequency @default. 3dmark11 is a bad CPU test, SuperPi 1M is too short. Only Fritzchess and Cinebench are meaningful. The slower Fritzchess result doesn't make sense. The pre-release non-professional testers are really pathetic nowadays. They should disable the turbo and should make sure the CPU runs at base frequency all the time. They can't handle the CPU Turbo or auto overlock features by some Z77 boards properly.

I'm well familiar with intel CPUs. MTed Turbo is not guaranteed by intel spec to be always at 3.9Ghz (in case of Haswell or 3770K).

I'm not saying the review is perfect, far from it. But there is no magic pixie dust left, sorry. This is Haswell and it's not that bad. It's just not major step up for SB/IB users unless one uses exclusively iGPU and doesn't care about discrete graphics. For pre-SB users it's a great step up.