Yonah sossaman Cinebench 2003 bench marks

[Intelia]

Very nice for early M/B and cpu . It looks like the yonah clock for clock is going to be all over the AMD 64's . It kinda looks like it could be real one sided. Merom and conroe should really be fast. Just imagine a Conroe @3.0 GHz than in 2007 ondie memory controller this is going to be good'

Cinebench 2003 (32 bit)
2x Dual Sossaman 2,0 GHz* 884

2x Dual Sossaman 1,5 GHz 663

1x Dual Athlon 64 X2 4800+ 638

1x Dual Pentium XE 3,2 GHz 610

1x Dual Athlon 64 X2 3800+ 536

1x Dual Pentium D 3,2 GHz 528
%%%%%%%%%%%%%%%
1x Dual Yonah 2,0 GHz* 521 AMD X-2 killer )
%%%%%%%%%%%%%%%
1x Dual Yonah 1,5 GHz 383

1x Single Dothan 2,13 GHz 286

1x Single Pentium XE 3,2 GHz 284

1x Single Dothan 1,73 GHz 228

1x Single Sossaman 1,5 GHz 206

http://www.computerbase.de/news/hardwar.../august/idf_benchmarks_sossaman_yonah/

 

[n7]

*sigh*

Why couldn't you keep your FUD in the other thread...

 

[CheesePoofs]

Yes, because clock for clock is the best type of comparison. :roll: A 2.8ghz A64 KILLS a 2.8ghz P4.

Edit: lemme guess: "OMG WE'RE NOT TALKING ABOUT AMD WE'RE TALKING ABOUT SOSSAMAN!!!"

 

[Avalon]

Let's assume that Intel's Yonah, if scaled up 20% in clock speed to match the clock speed of an X2 4800+, gained a 100% linear boost in speed in the process, shall we?

Cinebench 2003 (32 bit)
1x Dual Sossaman 2,0 GHz* 442

1x Dual Sossaman 1,5 GHz 331

1x Dual Athlon 64 X2 4800+ 638

1x Dual Yonah 2,4 GHz 625.2

1x Dual Pentium XE 3,2 GHz 610

1x Dual Athlon 64 X2 3800+ 536

1x Dual Pentium D 3,2 GHz 528

1x Dual Yonah 2,0 GHz* 521

1x Dual Yonah 1,5 GHz 383

1x Single Dothan 2,13 GHz 286

1x Single Pentium XE 3,2 GHz 284

1x Single Dothan 1,73 GHz 228

1x Single Sossaman 1,5 GHz 206
____________________________________________________________________


Aww, looks like AMD is better clock for clock than Yonah, and core for core than Sossoman.

 

[Intelia]

I have know reason not to expect 10%+ increase above that with a retail Cpu and motherboard . Also other Dothan cores scale up better than that . The higher the GHz the better they scale . Also ddr2 memory was running at 600Mhz only not 667 mgz.. Thats what yonah will have is a 667& FSB so add another 5%, It looks really good for intel. Bus speed was 2400 at 600FSB. Yonah runs at 2668bus speed at 667FSB . The actual no. will be much higher on a finished product.

I will get link showing dothan scaling up with higher ghz.

 

[Avalon]

Originally posted by: Intelia
Ya but I have know reason not to expect 5%+ increase above that with a retail Cpu and motherboard . Also other Dothan cores scale up better than that . The higher the GHz the better they scale .

You can't scale 5% above a linear scale. A theoretical 2.4 Yonah won't even hit the generous score that I gave it.

Nice link edit. Don't forget to mention the X2 3800+.

1x Dual Athlon 64 X2 3800+ - 536.
That's a 2ghz, 512KB L2 cache dual core processor, outperforming a FUTURE chip running in dual core at 2ghz with 2MB of L2 cache.

 

[CheesePoofs]

Originally posted by: Avalon

Originally posted by: Intelia
Ya but I have know reason not to expect 5%+ increase above that with a retail Cpu and motherboard . Also other Dothan cores scale up better than that . The higher the GHz the better they scale .

You can't scale 5% above a linear scale. A theoretical 2.4 Yonah won't even hit the generous score that I gave it.

ownage

 

[Furen]

Originally posted by: Intelia
1x Dual Athlon 64 X2 3800+ 536

1x Dual Pentium D 3,2 GHz 528
%%%%%%%%%%%%%%%
1x Dual Yonah 2,0 GHz* 521 AMD X-2 killer )
%%%%%%%%%%%%%%%

I'm sorry to break it to you my dear but the benchmarks show the lowest end X2 (which is available right now, by the way) beating the highest end Yonah that will be available in 6-8 months. The CPU is not an "early" cpu, by the way, those were the 90nm yonahs that were labeled as engineering samples. This is probably one of the first production trials, which, I assure you, will perform exactly the same as the final production ones. The chipset is just a respin of the one before it, so it should be pretty close to final performance as well.

Avalon, by the way, you cant just slice the scores for multi-processor CPUs in half because multi-processor scaling is not even close to linear (single to dual-core is an exeption to this since they scale remarkably well).

 

[Avalon]

Originally posted by: Furen

Avalon, by the way, you cant just slice the scores for multi-processor CPUs in half because multi-processor scaling is not even close to linear (single to dual-core is an exeption to this since they scale remarkably well).

I know, but I'm just following Intelia's logic here
Either way, I could just put it back to 2x Dual Sossomans and put what I think the estimated price for that setup will cost in brackets. But that would be sensible, and sensibility appears to be a color that Intelia is color blind to. Completely.

 

[Furen]

Originally posted by: Intelia
I have know reason not to expect 10%+ increase above that with a retail Cpu and motherboard . Also other Dothan cores scale up better than that . The higher the GHz the better they scale . Also ddr2 memory was running at 600Mhz only not 667 mgz.. Thats what yonah will have is a 667& FSB so add another 5%, It looks really good for intel

I will get link showing dothan scaling up with higher ghz.

OMG, she just raised it to 10%, beat that!

 

[Avalon]

Originally posted by: Intelia
I have know reason not to expect 10%+ increase above that with a retail Cpu and motherboard . Also other Dothan cores scale up better than that . The higher the GHz the better they scale . Also ddr2 memory was running at 600Mhz only not 667 mgz.. Thats what yonah will have is a 667& FSB so add another 5%, It looks really good for intel

I will get link showing dothan scaling up with higher ghz.

It doesn't matter if the Yonah that ships is 10% faster than the one in benchmarks. My scaling concept is giving the Yonah at a 20% mhz increase a 20% boost in performance, which never happens. If you were getting 50-75% of the scaling of my example, you'd be in great shape, but it still wouldn't be enough to tip the scale past AMD in a synthetic bench which looks like was run by Intel themselves.

 

[AnImuS]

Originally posted by: Avalon

Originally posted by: Intelia
I have know reason not to expect 10%+ increase above that with a retail Cpu and motherboard . Also other Dothan cores scale up better than that . The higher the GHz the better they scale . Also ddr2 memory was running at 600Mhz only not 667 mgz.. Thats what yonah will have is a 667& FSB so add another 5%, It looks really good for intel

I will get link showing dothan scaling up with higher ghz.

It doesn't matter if the Yonah that ships is 10% faster than the one in benchmarks. My scaling concept is giving the Yonah at a 20% mhz increase a 20% boost in performance, which never happens. If you were getting 50-75% of the scaling of my example, you'd be in great shape, but it still wouldn't be enough to tip the scale past AMD in a synthetic bench which looks like was run by Intel themselves.

Can i slap him/she now?

 

[Intelia]

Originally posted by: Furen

Originally posted by: Intelia
1x Dual Athlon 64 X2 3800+ 536

1x Dual Pentium D 3,2 GHz 528
%%%%%%%%%%%%%%%
1x Dual Yonah 2,0 GHz* 521 AMD X-2 killer )
%%%%%%%%%%%%%%%

I'm sorry to break it to you my dear but the benchmarks show the lowest end X2 (which is available right now, by the way) beating the highest end Yonah that will be available in 6-8 months. The CPU is not an "early" cpu, by the way, those were the 90nm yonahs that were labeled as engineering samples. This is probably one of the first production trials, which, I assure you, will perform exactly the same as the final production ones. The chipset is just a respin of the one before it, so it should be pretty close to final performance as well.

Avalon, by the way, you cant just slice the scores for multi-processor CPUs in half because multi-processor scaling is not even close to linear (single to dual-core is an exeption to this since they scale remarkably well).

It doesn't show that at all if you look at the FSB it was running at 600FSB Yonah runs @ 667 so that will take care of the clock for clock 2Ghz.Also A finished Yonah core will no dought run better also. Than there is the M/B itself A retail board will no dought perform much better.

 

[Intelia]

1x Dual Yonah 2,0 GHz* 521x 1x Dual Athlon 64 X2 3800+ 536 I have zero reason to believe that yonah running 667FSB won' beat the X2 3800 than A retail yonah and retail M?B will just increase that . I also have no reason to believe yonah won't scale as well as the 3800 does . If you clock a 3800 up to 4800 clock it should score the same as the 4800. or not. Not i guess I see there differant 1 has 1mb.cache the other has 512cache' Still the yonahs final no. will be better than they are showing here how much better will see soon enough as more leaks get out on to the web.

 

[AkumaX]

Originally posted by: Intelia

Originally posted by: Furen

Originally posted by: Intelia
1x Dual Athlon 64 X2 3800+ 536

1x Dual Pentium D 3,2 GHz 528
%%%%%%%%%%%%%%%
1x Dual Yonah 2,0 GHz* 521 AMD X-2 killer )
%%%%%%%%%%%%%%%

I'm sorry to break it to you my dear but the benchmarks show the lowest end X2 (which is available right now, by the way) beating the highest end Yonah that will be available in 6-8 months. The CPU is not an "early" cpu, by the way, those were the 90nm yonahs that were labeled as engineering samples. This is probably one of the first production trials, which, I assure you, will perform exactly the same as the final production ones. The chipset is just a respin of the one before it, so it should be pretty close to final performance as well.

Avalon, by the way, you cant just slice the scores for multi-processor CPUs in half because multi-processor scaling is not even close to linear (single to dual-core is an exeption to this since they scale remarkably well).

It doesn't show that at all if you look at the FSB it was running at 600FSB Yonah runs @ 667 so that will take care of the clock for clock 2Ghz.Also A finished Yonah core will no dought run better also. Than there is the M/B itself A retail board will no dought perform much better.

sooo.. where did you learn your engrish?

 

[Furen]

Originally posted by: Intelia

It doesn't show that at all if you look at the FSB it was running at 600FSB Yonah runs @ 667 so that will take care of the clock for clock 2Ghz.Also A finished Yonah core will no dought run better also. Than there is the M/B itself A retail board will no dought perform much better.

I dont know why I even bother with you... seriously.

First, the 600MHz FSB on that link is, of course, an error by CPU-Z. Second, the FSB on the yonahs is supposed to be 166MHz (which would be quadpumped so the "effective" Bus Speed--which is the speed right under the FSB in CPU-Z-- would be 667), Intel hasnt delivered any bus running at higher than 1066MHz did you think they suddenly were able to more-than-double the bus speed on a LOW POWER MOBILE? Third, if you dont know what you're talking about, please stop making an idiot out of yourself instead of trying to prove us wrong with your mindless speculation.

EDIT: one last thing, mobile boards are tweaked for low power not performance, I'd expect final boards to perform the same or worse, since all the power-saving crap will be enabled. If you doubt that the FSB is supposed to be running at 166MHz ask your dearest David to show you how fast it runs on his all-mighty dothan.

 

[TuxDave]

So I guess the bickering has shifted from Dothan vs AMD 64 to Yonah vs AMD X2. Anyone realize that in both of these arguments, they're not in the same platform? What's the power envelope of each anyways?

 

[Intelia]

Ya I found it they used DDR2 400 to do the test. So there is alot more performance there than what is shown

 

[JME Fidelity]

Wow, Intellia please contain your useless bullsht to one thread so other users may enjoy the forum and talk about things in an intelligent manner.

You get owned in one thread, so you decide to try to make another with benchmarks you made up?

Moron. DIAF

 

[Intelia]

Originally posted by: Furen

Originally posted by: Intelia

It doesn't show that at all if you look at the FSB it was running at 600FSB Yonah runs @ 667 so that will take care of the clock for clock 2Ghz.Also A finished Yonah core will no dought run better also. Than there is the M/B itself A retail board will no dought perform much better.

I dont know why I even bother with you... seriously.

First, the 600MHz FSB on that link is, of course, an error by CPU-Z. Second, the FSB on the yonahs is supposed to be 166MHz (which would be quadpumped so the "effective" Bus Speed--which is the speed right under the FSB in CPU-Z-- would be 667), Intel hasnt delivered any bus running at higher than 1066MHz did you think they suddenly were able to more-than-double the bus speed on a LOW POWER MOBILE? Third, if you dont know what you're talking about, please stop making an idiot out of yourself instead of trying to prove us wrong with your mindless speculation.

EDIT: one last thing, mobile boards are tweaked for low power not performance, I'd expect final boards to perform the same or worse, since all the power-saving crap will be enabled. If you doubt that the FSB is supposed to be running at 166MHz ask your dearest David to show you how fast it runs on his all-mighty dothan.

That was a picture of the 4 core sossaman of that CPU-Z and it clearly shows 4 cores running at 600mgz FSB each . 4x600 =2400 bus speed . I don't know y its showing it at the 2400 Bus speed It should be just 600 bus speed . It is running on ddr2 400 ram with a x2.5 multiplier

 

[Furen]

Originally posted by: Intelia
Ya I found it they used DDR2 400 to do the test. So there is alot more performance there than what is shown

OMG, SOSSAMAN uses DDR2 400 because it's a goddam SERVER CHIP and there ISNT ANY registered DDR2 667 (and registered DDR2 533 is as rare as your rational thoughts). Anyway, be asured that intel ran that Yonah as fast as they possibly could, they wouldnt let people benchmark things that are not running at their best unless they were a bunch of idiots, which they most definitely arent.

 

[JME Fidelity]

Originally posted by: Intelia

Originally posted by: Furen

Originally posted by: Intelia

It doesn't show that at all if you look at the FSB it was running at 600FSB Yonah runs @ 667 so that will take care of the clock for clock 2Ghz.Also A finished Yonah core will no dought run better also. Than there is the M/B itself A retail board will no dought perform much better.

I dont know why I even bother with you... seriously.

First, the 600MHz FSB on that link is, of course, an error by CPU-Z. Second, the FSB on the yonahs is supposed to be 166MHz (which would be quadpumped so the "effective" Bus Speed--which is the speed right under the FSB in CPU-Z-- would be 667), Intel hasnt delivered any bus running at higher than 1066MHz did you think they suddenly were able to more-than-double the bus speed on a LOW POWER MOBILE? Third, if you dont know what you're talking about, please stop making an idiot out of yourself instead of trying to prove us wrong with your mindless speculation.

EDIT: one last thing, mobile boards are tweaked for low power not performance, I'd expect final boards to perform the same or worse, since all the power-saving crap will be enabled. If you doubt that the FSB is supposed to be running at 166MHz ask your dearest David to show you how fast it runs on his all-mighty dothan.

That was a picture of the 4 core sossaman of that CPU-Z and it clearly shows 4 cores running at 600mgz FSB each . 4x600 =2400 bus speed . I don't know y its showing it at the 2400 Bus speed It should be just 600 bus speed . It is running on ddr2 400 ram with a x2.5 multiplier

Whats an mgz?

 

[Intelia]

So I think we can assume the sassaman was running at 100mgz for a total of 400mgz. Or would you say they got some fairly good O/C registered memory there. I know what its suppose to be running at I just wanted to make sure you knew it wasn't. So we agree at least that the sassaman performance is going to be a lot better. Unless you believe that registered DDR2 400 is running at DDR2 667 speeds. with a 2.5 multiplier its means its also running asyn correct. So with the right memory these things are going to shine at 1:1 with DDR2 667 memory aren't they. So now you got it so I completely understand. Boy it took you a while to get that threw my thick head didn't it LOL!!!
Of course the FSB can run at 667 but the fact that the memory is running at 400mgz or it really doesn't matter I thought the x2.5 multiplier was to keep the cpu running at 1.5 Ghz. I would suspect that yonah was also being held back why show your cards
. Yonah will perform better and sassaman a lot better. Still for a DP Dc settup I think its looking good DP isn't going to scale like DC . I think with the right memory and everthing properly settup I think we will see the score that sassaman 2x 2Ghz approach 1100 and I believe thats a good score.I mean 884 the way its set up 1200 wouldn't be pushing it would it . You know that the 400Mgz memory is really choking those 2 dual core CPU'S . Also we believe the memory is running at 400 so if a multiplier of 2.5 is used what would the FBS actually be running at ??????2.5x2=5 1500 divided by 5 =300x2 =600FSB

 

[Furen]

I'd say that the FSB bus is running at 667 and the memory is not running on sync with it, but rather is running at 400MHz, remember that the FSB is not just a link to the memory so even if the memory but also a link between the CPUs on the different sockets and a link to the rest of the system. If you look at the slide that has the DDR2 400 reference you'll see that they still have it spec-ed at 667. I'd accept that the Sossaman is not running under optimal conditions since there is no memory that can handle 1:1 with it's FSB but this is also revealed in the poor benchmark performance that it gives (and I'm sure even you'll admit that the sossaman performance sucks). All this, however, says nothing about Yonah.

 

[Furen]

I can understand Sassaman getting better performance but I still dont see where you get that Yonah will also get better. I, personally, think that Yonah is maxed out right now. In fact, I wouldnt expect it to clock higher than 2.26, which is great for a mobile chip but it wont hold a candle to the X2s to which you want to compare them (since they'll keep clocking higher and higher). Now, I'm not saying that the Yonah architecture cant beat AMD's X2, the problem is that yonah is a mobile architecture and, thus, is constrained by power consumption. Yonah's real "rival" will be AMD's Taylor, which I'd guess wont clock higher than 2.2 for a while (putting it right around Yonah's clock speed).