AMD shows sunny side of Barcelona

[r0ck]

http://news.com.com/AMD+shows+sunny+side+of+Barcelona/2100-1006_3-6185526.html

The quad-core chip processed about 4,000 pixels per second in rendering the image, while the dual-core chip could only hit around 2,000 pixels per second. Allen said Barcelona was not running at the fastest clock speed that will be available at launch, although he declined to specify the speeds that will be available.

AMD did not test Barcelona's performance against one of Intel's quad-core Clovertown Xeon processors, but Allen said Barcelona "will be the highest-performing x86 chip out there. It will blow away Clovertown."

http://www.2cpu.com/review.php?id=114&page=11
http://www.realworldtech.com/page.cfm?ArticleID=RWT111406114244&p=2
At 2.33GHz, Clovertown is near 4000...

 

[xpresso]

one benchie at a time

 

[ches111]

I hope it smokes...

Need MORE Benchies..... Independent Benchies!!!!!

Need more data....

Need more input...

Release the thing already

 

[MU_Engineer]

r0ck, the 3819 px/sec figure is for two 2.33 GHz E5345s while the news.com.com's story had one Agena hitting 4000 px/sec. The clock of the Agena wasn't given, but I'd hazard a guess that it wasn't that far away from the 2.33 of the E5345. Pov-Ray has been noted to scale pretty well, so the Agena may very well be twice as fast per clock at Pov-Ray than the Intel quad-cores. Very interesting.

 

[halbhh]

well...there are certainly a lot of....interesting....possibilities. It's a fun time for watching it all unfold.

 

[turpit]

@ Rock....OMG almost a real K10 benchmark!!! Its thin, but I guess its the best we're gonna get for now, it looks good, and they used a real benchmark (POV-Ray) instead of "simulations or applicaions". Lets hope it loks as good on other progs

Ahh I see you got in on dasickninja appreciation week too. :lol:

aye turpit invited me. I answered with an "ahhh for fux sake" and he elaborated with a good argument.

Johan joseph fux wrote the first treatise on counterpoint in the 1700's

http://www.baroquemusic.org/bqxfux.html

Yea turpit invited me as well.


Ya know, you would think Ches111 would catch on and join in, but since he hasent....

@Ches111
YO Ches... its Ninja Tribute week...were all giving 'tribute' by sporting a genuine 'DaSickNinja' avatar...how about joining the trend

 

[ajfink]

r0ck, the 3819 px/sec figure is for two 2.33 GHz E5345s while the news.com.com's story had one Agena hitting 4000 px/sec. The clock of the Agena wasn't given, but I'd hazard a guess that it wasn't that far away from the 2.33 of the E5345. Pov-Ray has been noted to scale pretty well, so the Agena may very well be twice as fast per clock at Pov-Ray than the Intel quad-cores. Very interesting.

I agree. It could easily have been a 2.3Ghz model or a 2.5Ghz model. I wish they would have given the clockspeed.

 

[dasickninja]

Okay, a benchmark against itself. How enlightening.


On another note, I have trouble finding myself. *Looks at the many DaSickNinja's running around*

 

[r0ck]

r0ck, the 3819 px/sec figure is for two 2.33 GHz E5345s while the news.com.com's story had one Agena hitting 4000 px/sec. The clock of the Agena wasn't given, but I'd hazard a guess that it wasn't that far away from the 2.33 of the E5345. Pov-Ray has been noted to scale pretty well, so the Agena may very well be twice as fast per clock at Pov-Ray than the Intel quad-cores. Very interesting.

The company ran a demo comparing the performance of two four-socket servers, one using the quad-core Barcelona chip and one using a dual-core Opteron chip. The demo measured the performance of the chips on an imaging benchmark called POV-Ray, and as you might expect, the quad-core chip finished its task quicker than the dual-core chip.

That's strange, why would they bench a single chip in a four socket setup? Twice as fast as Clovertown in POV-Ray? If that was so, AMD could have made good PR and crushed Clovertown on the spot. I think news.com is off somewhere. We should soon hear from other sites.

http://www.hothardware.com/articles/AMD_QuadFX_Platform__FX70_Series_Processors1/?page=10

Or maybe not so far fetched, with a single QX6700 getting 2702 here...

But assuming that one quad does get 4000, do you have any idea why? I don't think the app is SSE optimized as AMD is competitive (relative to most others) and it's not memory intensive (near perfect scaling even on FSB).

 

[dasickninja]

*Bows graciously to the Forumz*

 

[MU_Engineer]

Even then, 2660 vs. 4000, it's still a pretty major difference. The thing that I want to know is if the Pov-Ray version they used supports SSE4a. If it does, then the Agena performance will be much, much better in comparison to the Core 2 Quads than it will be in everything else without SSE4a. And the 45 nm Core 2s have SSE4, so that will even things out a lot more. However, I don't think that Pov-Ray has any alpha builds with SSE4a out there so it might be that Agena is just that fast without any "special enhancements," at least in Pov-Ray. But this was a specific benchmark touted by AMD, so the 50% gain over the Kentsfield probably is one of the biggest advantages it will have over the Kentsfield.

I highly doubt that this was a 16-core, 4-socket test bed as a dual-socket 2.4 GHz Opteron 280 did 1711 px/sec. It should certainly be possible to get 4000 px/sec using four of the faster current dual-core Opterons like the 8220 and 8222. 4000 px/sec from twice the number of enhanced cores is a SERIOUS PROBLEM. Also, the article mentioned that the dual-core part did 2000 px/sec. AMD is not going to make any dual-core 10h server parts, so that chip had to be a single-socket model. And if a single dual-core chip did 2000 px/sec, then a quad should do about 4000. The numbers make sense in relation to each other. A four socket rig should do about 15k px/sec if the numbers are right.

 

[tamalero]

*Bows graciously to the Forumz*

one.. of..us.......

 

[turpit]

Okay, a benchmark against itself. How enlightening.


On another note, I have trouble finding myself. *Looks at the many DaSickNinja's running around*

Cant find the trees for the forest?

 

[the_vorlon]

r0ck, the 3819 px/sec figure is for two 2.33 GHz E5345s while the news.com.com's story had one Agena hitting 4000 px/sec. The clock of the Agena wasn't given, but I'd hazard a guess that it wasn't that far away from the 2.33 of the E5345. Pov-Ray has been noted to scale pretty well, so the Agena may very well be twice as fast per clock at Pov-Ray than the Intel quad-cores. Very interesting.

I believe you may be misreading the article actually...

The company ran a demo comparing the performance of two four-socket servers, one using the quad-core Barcelona chip and one using a dual-core Opteron chip.

I believe this was a result with a 4 socket machine?

 

[r0ck]

r0ck, the 3819 px/sec figure is for two 2.33 GHz E5345s while the news.com.com's story had one Agena hitting 4000 px/sec. The clock of the Agena wasn't given, but I'd hazard a guess that it wasn't that far away from the 2.33 of the E5345. Pov-Ray has been noted to scale pretty well, so the Agena may very well be twice as fast per clock at Pov-Ray than the Intel quad-cores. Very interesting.

I believe you may be misreading the article actually...

The company ran a demo comparing the performance of two four-socket servers, one using the quad-core Barcelona chip and one using a dual-core Opteron chip.

I believe this was a result with a 4 socket machine?

That is the confusing part though.

 

[Dade_0182]

Nice find and for once the info didn't come from the inq or fudzilla. I've just heard too much from them this last month or so...

 

[NMDante]

Neat.

That is a lovely box, I must say. Looks pretty big. Not sure how much weight it can actually hold, but it does say Entertainment Center on it, so I assume it will hold at least 10 lbs.

 

[badge]

Neat.

That is a lovely box, I must say. Looks pretty big. Not sure how much weight it can actually hold, but it does say Entertainment Center on it, so I assume it will hold at least 10 lbs.

Weighs 45 lbs.

I've had the Mozart case for many months. Have the Thermaltake Media Center remote control panel too. Have not found the Thermaltake 7" retractable LCD for sale anywhere. The case is extremely heavy. I have eleven 120 mm fans running. The best cooling solution I have had by far.

 

[badge]

I have a circular/L shaped computer desk station, an older one. The Mozart case was too tall to slide under, so I had to put blocks under the table on both sides and the middle to fit the case under the table. Now my '67 GTO is out front flat on the lawn.

 

[badge]

I have a circular/L shaped computer desk station, an older one. The Mozart case was too tall to slide under, so I had to put blocks under the table on both sides and the middle to fit the case under the table. Now my '67 GTO is out front flat on the lawn.

My dad had a 67 gto.He used to race it in stock car eliminations Did they stop making blocks?

Kuhl car. A buddy of mine had a '67. We used to race and see who could be the first to replace the transmission for the sixth time. I won a trophy on the quarter mile one day. I had a ragtop '64 Chevy 2 door with SNOW TIRES. Only car in my class. The track announcer said, "If that Ragtop is coming back next week, He'll have to get some new doors 'cause He just got 'em blown off!" I still have the trophy.

 

[badge]

A grade scool friend of mine took super street eliminator several times when we were in high school. he took me for a ride one night over the old neighborhood. When he hit third gear my head hit the roof and I landed on my jewels. He 'bout killed me!

 

[badge]

Nice. We only had a pinto to race around in when i was a kid.Not head slamming but fun enough on the right curves with the right speeds. :lol: We didnt tell my dad about the bent rim for decades.

A Pinto. Al Franklin had a Pinto. he put a header on it and it sounded just like an electric razor. Then he took out to the off road motorcycle jumps. It was his mom's new car. When he got about 5500 miles on it, my buddy Gary Browne dared him to jump the tracks. Al did. he bent the frame so bad they had to junk it! Al served some time later, but so did a lot of guys I know.

 

[badge]

An old buddy of mine who incidently just finished 4 of a 12 year sentence (LOL!) used to deal cars. he had a classic saying I say to this day. "I shined the wheels, that jacks the price!"

 

[easyg]

The company ran a demo comparing the performance of two four-socket servers, one using the quad-core Barcelona chip and one using a dual-core Opteron chip. The demo measured the performance of the chips on an imaging benchmark called POV-Ray, and as you might expect, the quad-core chip finished its task quicker than the dual-core chip.

Vorlon, you are correct, they did two 4P setups, one with quad barcey and one with dual optys, if this is the case, then this is not that impressive... there is something amiss or the clock delta was large....

Again, too bad we did not have clock speeds to compare -- as such, the scaling is meaningless.

jack

Indeed. More meaningless numbers. I remember reading in George Ou's blog that even the scanty info AMD has released about K10 simply doesn't add up. Almost like a deliberate misinformation campaign.

 

[WR]

Substantiating with info gleaned from the Inquirer article:

All four sockets were populated on the development systems, yielding 8 K8 cores and 16 K10 cores. AMD claimed they were using 65W TDP chips - thus the benches don't compare to top-of-the-line clocks, but the two systems should use the same amount of energy.

So performance-per-watt nearly doubled going up a generation, a good sign itself, but I'm concerned because Clovertown already holds much better PPW than Opteron, and Clovertown is older technology than either K10 or Intel's upcoming 45nm shrink.