i7 4770k new 3dmark physics test

[csbin]

:whiste::whiste:

 

[Face2Face]

So it that good? How does it compare to the 3770K?

 

[Red Hawk]

Wait, 4770k? Is this an Ivy Bridge refresh or rebadge?

 

[cytg111]

haswell right?

Quick, get some scores

 

[inf64]

Physics score is relevant here since it measures x86 performance, but GPU score is interesting.
For GPU comparison here is 5800K:
http://www.3dmark.com/cg/139753
Here is 5800K with 7700 series hybrid CF:
http://www.3dmark.com/cg/113537

For x86 compression here is 4.4Ghz 3770K's result in physics:
http://www.3dmark.com/cg/57720
At 3.5ghz(a clock it would most likely run at in this benchmark) the 3770K would score ~6710pts. 4770K @ stock scores 7053pts. Both are 3.5/3.9Ghz parts(supposedly true for 4770K).
Maybe IDC can run new 3dmark on his machine if he finds time so we can know what it gets in Physics subtest @ stock.

 

[cytg111]

"
At 3.5ghz(a clock it would most likely run at in this benchmark) the 3770K would score ~6710pts. 4770K @ stock scores 7053pts. Both are 3.5/3.9Ghz parts(supposedly true for 4770K).
"

- That is a little .. little :/

 

[DigDog]

delid! delid!!

 

[inf64]

"
At 3.5ghz(a clock it would most likely run at in this benchmark) the 3770K would score ~6710pts. 4770K @ stock scores 7053pts. Both are 3.5/3.9Ghz parts(supposedly true for 4770K).
"

- That is a little .. little :/

Yes but we need a real 3770K@ stock score . Then we can compare them as it should be done.

 

[shady28]

"
At 3.5ghz(a clock it would most likely run at in this benchmark) the 3770K would score ~6710pts. 4770K @ stock scores 7053pts. Both are 3.5/3.9Ghz parts(supposedly true for 4770K).
"

- That is a little .. little :/

1% better than a similar clocked 3770k...

That's probably well within the margin of error for these benchmarks too.

 

[Haserath]

Memory speed usually affects physics tests quite a bit. Can't tell how much better it is with this.

 

[MisterMac]

2-3% IPC increase?

Ouch.


That's even worse then the previous "leak".



Hope this doesn't turn into bulldozer part deux for intel :C

I want to upgrade with my tax dollars!

 

[frozentundra123456]

1% better than a similar clocked 3770k...

That's probably well within the margin of error for these benchmarks too.

That is a difference of 340 out of 6710 or slightly over 5%. How did you get 1%?

 

[tweakboy]

Guys on Q3 2013 Asus will unveil its Ivy Bridge E processors.

6core and 8core desktops and 10 and 12 core xeons E

I just upgrade BIOS and slap on a 8 core. cant wait 16 threads wow.

 

[Termie]

:whiste::whiste:

Here's my 3770k locked at 3.5GHz:

Physics Score 6840

And here's my 3770k at 3.7GHz, which is what a stock x770k chip should run at under full load:

Physics Score 7204

I'd say there's not much improvement, if any, in that 4770k score. Assuming it's locked at 3.5GHz, it's a 3% IPC improvement.

 

[tweakboy]

WTF is a 4770k BTW Ivy Bridge E processor 2013 Q3. These processors will blow away anything Haswell has for 2013 at least.

Im gonna grab a 8 core, 16 threads , wow,,,,,,, How many core is that 4770k hmmm.

 

[tweakboy]

Ivy Bridge E in 2013 second half. Models

• Core i7-4930
• Core i7-4960
• Core i7-4970
• Core i7-4990

 

[DigDog]

all they really need to do to improve drastically is change the goop that's between lid and die

 

[inf64]

Here's my 3770k locked at 3.5GHz:

Physics Score 6840

And here's my 3770k at 3.7GHz, which is what a stock x770k chip should run at under full load:

Physics Score 7204

I'd say there's not much improvement, if any, in that 4770k score. Assuming it's locked at 3.5GHz, it's a 3% IPC improvement.

Thanks for running the test on your system.
In this particular test there seems to be very small improvement from IB to Haswell. It's rather strange since Phycis is well MTed and additional load port and exec. unit should be able to at least provide some benefit. IB is well rounded core it seems , intel is nearing the edge of extractable IPC in legacy code. If this Physics subtest was running some specially coded AVX or FMA(if applicable) code then Haswell would probably storm ahead by some healthy margin. This is what makes it more "future proof" core.

For regular Joe users ,difference between IB and Haswell will probably be unnoticeable except when more powerful integrated GPU is used of course.

 

[moonbogg]

No way to know if true, but if it is, then my rig is going to last a really REALLY long time.

 

[YBS1]

No way to know if true, but if it is, then my rig is going to last a really REALLY long time.

Yeah, 4 digit physics scores are cute.

 

[mikk]

Here's my 3770k locked at 3.5GHz:

Physics Score 6840

And here's my 3770k at 3.7GHz, which is what a stock x770k chip should run at under full load:

Physics Score 7204

I'd say there's not much improvement, if any, in that 4770k score. Assuming it's locked at 3.5GHz, it's a 3% IPC improvement.

Did you run with iGPU? If not, your score is not comparable. GPU has a small effect on physics scores.

 

[Termie]

Did you run with iGPU? If not, your score is not comparable. GPU has a small effect on physics scores.

No I didn't. How would running with integrated graphics affect a measure that is clearly a CPU test?

 

[frozentundra123456]

Here's my 3770k locked at 3.5GHz:

Physics Score 6840

And here's my 3770k at 3.7GHz, which is what a stock x770k chip should run at under full load:

Physics Score 7204

I'd say there's not much improvement, if any, in that 4770k score. Assuming it's locked at 3.5GHz, it's a 3% IPC improvement.

I can't even see a source for that graphic by the op. Am I missing something?

 

[Zink]

RAM bandwidth getting eaten up by the iGPU?

 

[grimpr]

Kaveri is gonna wipe the floor with it.