Next Gen VS. Current Gen CPU: Crysis Benchmark

[knightwolf654]

So how do you explain the big spread in performance between the various CPUs?

Any game benchmark that removes the GPU as a bottleneck is a 'good' CPU benchmark. Not 'real world' but it does show the potential of the CPU.

optimized for core 2
loose memory timings

[Hornet331]

cookerjc any chance you can run 3d05/06 and show the cpu score, to stop this mad babbling?

[informal]

I still can't access the page linked in the news post.I get this:

Not Found

The requested URL /news/hard/2007-11-05/1194231291d6785_1.html was not found on this server.

Did anyone save the graphs and is it possible to attach it as image here?

[Hornet331]

no problem here, i guess your using firefox.

[informal]

Nope,same thing i get in Opera and IE.Strange.Must be my ISP messing around again.Let me grab my assault rifle,brb .

[RPGWiZaRD]

optimized for core 2
loose memory timings

Yea I agree but there are also other factors, depends how they tested it too but at lower res I'm sure you will see a difference of course but not necessarily as much difference it could be in an optimal scenario. But I take back the word "indicator" cuz every game at low res is a good indicator at CPU performance but still not an optimal CPU test IMO especially the GPU heavier they get and Crysis demo is one good example.

Below you see how I prefer making a CPU performance comparision. One is with my AMD CPU @ 2.8GHz, the other is with Intel @ 3.75GHz, both used my 7900GTO at same speed in this test. I don't think it's necessary to tell which screen is with AMD and which is the one with the Intel CPU...

If you want I could also take a screenshot when running the Intel CPU at stock speed for comparision's sake, it should be around 41% FPS advantage at the overclocked speed though (3.75GHz / 2.66GHz =~ 41%).

[NickS]

How come they lowered the HT FREQ so much from 2915 to 3GHz? RAM dropped too.

[Epsilon84]

optimized for core 2
loose memory timings

So any game that is 'optimised' for Core2 isn't a valid benchmark now? Too bad Intel has nearly 80% of the market, so the vast majority of CPU sales happen to be Core2 based, geez Crytec are so biased optimising for the majority instead of the minority! Welcome to the real world dude.

I've already acknowledged the memory timings issue, yes, they are crap, AMD will do better with better timings, but so will Intel, but not to the same extent obviously. Still, no amount of memory tweaking will get Phenom anywhere NEAR Core2 in this benchmark, and you know it.

I'd still say that, from these numbers, it's a darn good CPU benchmark, performance increases from faster memory as well as clockrate/core count.

It is much better than the last one showing almost identical numbers between the E6850 and QX9650, at least this one shows some dual core -> quad core scaling.

How come they lowered the HT FREQ so much from 2915 to 3GHz? RAM dropped too.

If you open the link, you will find more detailed tests done where the HT/RAM speed is increased, there is also a 4-4-4-12 timings test, I think you have to click on the images to access those results though.

To sum up, yes, faster memory and lower latency helps, but it still only gets the Phenom to ~71fps @ ~3GHz, compared to a completely stock and 'untweaked' C2Q 3GHz @ ~81fps. I'm sure you can easily get the C2Q to ~85fps by increasing the FSB/RAM and tightening memory timings. So yes, the gap does get smaller with better timings, but only marginally. Of course, this is before the C2Q gets overclocked to 4GHz+... I'd imagine things will get pretty ugly then... but thats another topic altogether isn't it.

[Lastviking]

Hmm i wonder how many cores crysis use.... , and i also dont get it why they cant compare same speed on the memmorys(it looks very bad way to compare K10 vs another cpus)

[AbelJemka]

I don't understand the difference between old test and new test
Who can explain me?
old test : http://news.expreview.com/2007-10-29...0532d6599.html

[realsmasher]

this is bull.

i can get 110fps avg with my E6600 @ 2,4ghz and details and res @ low, so this cpus here MUST get more or this is all a GPU-limit.

so if they want to compare the CPUs, they better choose a situation where the gpu does not matter.


is this realistic ? No, but a cpu benchmark with crysis is not useful at all.

It's like trying to break the world-speed-record with your car while driving through london.

[cookerjc]

I don't understand the difference between old test and new test
Who can explain me?
old test : http://news.expreview.com/2007-10-29...0532d6599.html

because we find out too much variables in the first test, so we re-install the OS, using the stable driver, and do it again...

[AbelJemka]

What are thoses variables which boost up the score by more than 40%

[cookerjc]

What are thoses variables which boost up the score by more than 40%

a soundcard driver or something else
anyway, the problem solved..

sorry to all ppl that can not access to our pages.
Today we change our host so there maybe some ip translate problem..

[GoThr3k]

is it just me,or is that site really really really slow

[AbelJemka]

Impressive what a bad sound driver can do
I may update mine i think

[RaZz!]

How come they lowered the HT FREQ so much from 2915 to 3GHz? RAM dropped too.

me was thinking the same. those settings seem kinda unusual... i wouldn't even run my 4000+ with such odd settings

[madcho]

HT3 is off ...

[[XC] hipno650]

I found this interesting topic over at techpowerup forums.It considers the low min. fps AMD Phenom had in earlier news posts(it was around 8.6fps if i recall correctly).
Now look at this post from a C2D user with a CPU clocked way up to 3.6Ghz:
http://forums.techpowerup.com/showthread.php?p=509527

Minimum fps is at the almost exact frame and it is 8.89 fps !
So C2D ,we saw ,has almost the same scores as C2Q(Penryn based too) and in this benchmark this user got the exact low minimum fps as Phenom got in expreview benchmark..What gives??



Now look at this post in the same thread :
http://forums.techpowerup.com/showpo...4&postcount=20

The user LiNKiN is running a Dual Core(Toledo) Athlon X2 CPU @2.85Ghz and he is getting min. fps in CPU test locked at 38fps?? The trick is in the CPU/Med/Low/High settings it seems.The user than posted another set of screens showing that minimum frames per second heavily depend on the setting used in CPU benchmark!
Screenshot(min 38fps/Low):

Look at the other shots and see for yourself how the minimum fps is all over the chart(depending on the settings in the demo)!


More ,this time with heavily OCed C2D@3.7Ghz scoring 10 fps as minimum fps in CPU test ,here:
http://forums.techpowerup.com/showpo...3&postcount=28

So we have now one Phenom@3Ghz having a minimum frame rate of ~8fps,one C2D@3.6Ghz having the same min. fps,one Toledo @2.85Ghz having 36.5fps in the same test,and we have a range of intel C2Ds and C2Qs @ expreview which have also higher than 20fps in minimum segment of the CPU test.It seems it is very dependent on the in-demo setting used,no matter if you ran CPU or GPU "benchmark"!

Anybody else seeing how "problematic" this "benchmark" is?
I am still looking into the "GPU" part of the tests,trying to see if there are also some discrepancies...

you know how the benchmark works right? it uses the settings you have set in the game (res, detail, physics etc.) your comparison form 2 different people with obviously two totally different set ups and resolutions just can't be done. look at the maximum and average FPS as well. the AMD has way higher numbers than the core 2 even thought the core 2 is way faster in game. this is why the numbers are so off: the person with the core 2 is running a high res with high detail levels and maxed out physics level. the person with the AMD is running everything on low at 800X600. thats why the AMD has a higher FPS. heck even my pentuim D would kill the core2 if i ratcheted down my detail and ran on 800X600.

on topic:
Phenom looks to be way behind. and it's just plain sad now. what more do we need to see? we have PI we have game benches. and they all tell the same story. K10 is slower than intel clock per clock. thats the hard reality people. and no ram timings is going to narrow that 25% gap. the only way phenom can survive as a good CPU and be competitive is sell for dirt cheap.

which means lots of Quads for me

[SADS]

i was also surprised when the HT was lowered, if you look the Phenom gains 100mhz over those 2 tables yet scores lower in the one with the lower HT even though it has an extra bit of clock speed.

perhaps the CPU is held back quite a bit by the memory and HT speeds?

[SEA]

I see another crap measurment from those guys...

1) why
Phenom X4 @ 2915MHz, mem @485 results 67.795
Phenom X4 @ 3000MHz, mem @375 results 64.325
Apparently X4 suffers from memory bandwich here: incorrect setup used - big timings (and wrong memory controller mode in bios?).

2) last table shows that there was just incorrect measurement somewhere:
Phenom X4 @ 2915MHz, mem @485 results 67.795
Phenom X4 @ 3048MHz, mem @508 results 71.9425

So, CPU overclock 4.56% togerther with Mem overclock 4.74% results 6.11% gain ((71.9425-67.795)/67.795 = 0.061177078) ? No way!

Overclocing only CPU or only Mem by X% shoould result in < X% gain,
and only proportional overclocking of all parts by X% can give X% gain. But not 1.5X % !

[madcho]

An overclock on a intel system of X&#37; give you <X% of performance.

The phenom can have >X%, this is the only one cpu witch can do this.

But i really think this is a fake.

[Soulburner]

Guys, a couple of things.

1. The Crysis timedemos run with the settings that you set in-game the last time it was run. So if you are running on all low the benchmark will be run at low settings. This makes it very hard to compare to a timedemo run between two different users.

2. Crysis is not a good CPU benchmark. CPU timedemo results will be affected by changing GPU's - my average CPU framerate went from 27 to 35 by changing my GPU from an 8800GTS to an 8800GT.

[SEA]

Ok.
So if we cannot compare 2 different results of X4, we cannot compare anything else - neither X4 to yorkfiled nor X4 to kentsfield nor yorkfiled to kentsfield.

[Clairvoyant129]

Not the best benchmark for CPUs but it does show that C2Qs outperform Phenom X4 by ~25%. The timing on AMD is crappy but whatever the memory timings are, it won't cross the 25% gap between the two.

With these early indications, I would have to guess that Phenom X4 is 20%-25% slower clock for clock.