My SB750, HD4850 x-fire, 9850, H20 Build...

[informal]

Well it's intel's compiler,they made it the way it is.Software companies can use other compilers out there,but most firms prefer intel's since it gives the best performance on intel's chips(which are the most spread CPUs on the market).
It would be nice to see a Linux open source test bench application suit(like the one Phoronix use),which could be compiled with best compilers out there for both intel and AMD.Crippling the AMD to less optimized SSE path(as shown before in my previous post) has serious performance effects on AMD's chips.I wonder what paths are used by such apps and games as Lost Planet,FutureMark Vantage,Cinebench etc. when you run them with non-intel CPUID.

[Brother Esau]

Right as this would come as a Big Surprise to us all

[Boschwanza]

performance rigging?

didnt know, honestly. its a shame a company such as intel would do such a thing

or is it the software companies?

No thats just Intel, but of course they can say .. Hey, thats our compiler why optimizing for other brands.

Nevertheless the Intel Compiler and the Dispatcher are one point of the Anti Trust lawsuite against Intel (at least here in europe). But thats becoming very off topic now. Sorry for that.

[kazgirl]

Cinebench 32 Bit realy sucks with the phenom. Bad Scaling etc.

Try 64 bit and try to get NB as fast as possible and you will get an additional extra boost in performance by near as 10 %

64-bit and 2.4Ghz NB... 9850Be @ 2.8Ghz

For comparison only... Rendering (xCPU) = 11207 edit: (pic is a bit small sorry..!!)

[bingo13]

I'd be very interested in this bios too Bingo... I've Googled but can't seem to find it anywhere.

Could you get us a link?
Thanks.... Dave

I will get it and the new Gigabyte BIOS uploaded today.

[Particle]

I will get it and the new Gigabyte BIOS uploaded today.

Yes! Where will they be uploaded to though, exactly?

[keithlm]

Yes! Where will they be uploaded to though, exactly?

He can't tell you that. It's secret. But rest assured: They are uploaded.

:-P

(Actually I want to hear your results with the NEW bios. I want to hear you say: "Oh yeah.")

[Rammsteiner]

My best with the 9600BE was 2.6ghz 100% stable with stock v-core. My 9600 hated voltage..2.7 was worse then 2.8 actually lol...

I actually meant your 9850BE

[Aquineas]

Cinebench is highly Intel optimated thats not new, at least for 64 Bit operating System Cinebench is forced to use SSE2 on every processor.

Well the Athlon 64 and up support SSE2, so AMD should be fine there, UNLESS they're using the Intel compiler. According to this post, the Intel compiler doesn't check for feature flags in the CPU to make the determination whether the CPU supports SSE2 or not, but instead relies directly on CPUID, which would basically mean that any SSE2 code would not be executing on the AMD cpus.

Thats one explanation the Phenom does much more better with 64 Bit in Cinebench except the better 64 bit implemetation on AMD CPUs

For example (Cinebench)
32 to 64 Bit on a Intel system gives you an extra boost about 10 %
32 to 64 Bit on a AMD system gives you an extra boost about 20 %.

Interesting, I didn't realize there was a difference in the 64 bit implementations. I suppose I could test it; I'm running 4cpus at work (2x Core2 Duo machine) in Windows Server 2003, and have both a Phenom at home running Vista 64 as well as a Athlon 4800 running xp64. I could build something using Visual Studio on both (I don't have a license to Intel, sorry). The question becomes, what would be a good single-threaded open source benchmark to compile..

Sorry, this represents a thread hijack. I'll back out now :-).

[charged3800z24]

I actually meant your 9850BE

sorry...

[Particle]

So was the secret thing humor or serious? I can't tell for sure. In the words of a Hick Uncle, "I needs me some biooos!"

[charged3800z24]

well this first release eith hates me or they need to update this bios. I can't change NB or HTT with out a hang on boot...I also have to run my ram at 800mhz or I can't OC..wierd...oh well

[charged3800z24]

so far 3.0ghz...

[Boschwanza]

Preety poor Score . For an extra boost scaling is very important and depends on the NB Speed, as far as i figured out NB Speed should be as close to the cores as possible. Final score should be something around 12.000

[charged3800z24]

Preety poor Score . For an extra boost scaling is very important and depends on the NB Speed, as far as i figured out NB Speed should be as close to the cores as possible. Final score should be something around 12.000

I set the NB to 2200mhz and my mem back to 1066mhz.... the score was actually less... I ran it three times.. averaged 100points lower.. I wil tighten my mem timings, but don't think it will help..

[charged3800z24]

Phenom 3.0ghz, HD 4850 (I had to RMA the other 4850..will update when it arives.)

[SocketMan]

Didn't AMD sue Intel a few years back just because of the
Compiler issues?

[keithlm]

So was the secret thing humor or serious? I can't tell for sure. In the words of a Hick Uncle, "I needs me some biooos!"

I was totally not serious. Sorry.

[charged3800z24]

Win 64 is not liking anything higher then 3.0ghz... now thi better then my ol board where 2.9 was not stable. 3.0ghz has been running all night great. I am going to load vista and try a vantage run at 3.2ghz if i an.

[Boschwanza]

I set the NB to 2200mhz and my mem back to 1066mhz.... the score was actually less... I ran it three times.. averaged 100points lower.. I wil tighten my mem timings, but don't think it will help..

Just made some few test figuring out that Cinebench is very inconsistent in scoring. It all depends on how the cinebench scheduler actually tasks the cores (thats different in every run and you can see it during the test)

So i get scores from 9466 to 10405 with exactly the same Settings which means a delta of 10 %.

[Rammsteiner]

So i get scores from 9466 to 10405 with exactly the same Settings which means a delta of 10 %.

Right wtf.

Another benchmark not to pay attention to then. Damn, not much left now.

Anyhow, charged, could you try to OC with HTT instead of multiplier? Just curious if the HTT does better now. Also try to run your memory 1:1 and some lower timings, see if if that does better. Ive my doubts regarding performance with Agena+8:3. It hardly allows tighter timings than tCL5, unless you get the HTT to 266Mhz and run the RAM 1:1, then it will.

Unless you can pull off ~DDR1200 with 8:3 and HTT of 225Mhz with the tightest timings as possible,8:3 is not worth it IMO.

[Brother Esau]

How are these boards doing guys?

[Particle]

In short, everyone but me has had good results. Some people are getting 200-400MHz increases. I on the other hand didn't move a single MHz. I'm still waiting on some sort of mystery BIOS for my ASUS though, so things may change.

[Brother Esau]

What processor do you have Particle? If you may be interested I am going to sell my 9850BE which has never been beyond clock and voltage in my sig due to my cats illness and time constraint for that reason.

[Particle]

I'm on a 9850BE week 10.