Bandwidth challenged Athlons

[Alex99]

Intel P4 runs at 400 FSB, and next yr Intel plan to boost that up to 533. Athlons are still stuck at 266, even the Palominos. Does anyone know if/when AMD would improve on this? Apparently the Nvidia nforce chipset could pump out 4.2 G/s, but Athlons FSB is the bottleneck. Today's multimedia intensive apps, like MPEG2/MPEG4 encoding/decoding heavily relys on memory bandwidth. The P4 clobbers Athlons in this area. I'm a big Athlon fan, but I also do alot of MPEG stuff. I am looking at maybe going with a P4 system. I know I'm gonna get yelled at, but my primary question is does AMD have plans to improve the Athlon's memory bandwidth?

[AlanSr]

I agree AMD needs to improve their design. I hope they will do something quick because I'm a huge AMD fan and I don't want to see them fall.

[UnseenMenace]

Bottom line is that Pentium 4 is indeed able to deliver excellent performance once software has been optimized for it. The Athlon's inability to beat Pentium 4 in this suggests that Athlon runs into bandwidth limitations with this streaming benchmark. In case of MPEG4-encoding Pentium 4's fast quad-pumped 100 MHz bus plus the dual-Rambus channel memory access of i850 seems superior to Athlon's dual-pumped 133 MHz bus and the 133 MHz DDR-SDRAM memory solution of AMD. That's why even the best optimizations can't give Athlon enough of a boost to overtake Pentium 4.

It is worth remembering that most of the information shown proving this very case happened to use the FlasK MPEG 4 that showed a very different behavior after Intel optimized its executables. The publication of the new benchmark results scored with the Intel-optimized version of FlasK was supposed to show how unreliable benchmarks can actually be. Especially a time-critical application such as a video encoder is able to twist around benchmark results several times. Wait what will happen once we've received the final Athlon-optimized executables!

Intel has used video encoding to show off Pentium 4's performance on numerous occasions before. We know now that those presentations are just as questionable as the initial results scored with FlasK.

What we learn from the FlasK MPEG issue can be translated to any other application. Once it was optimized for Athlon, Pentium III or Pentium 4 it will run particularly well on the according processor. The architecture of today's CPUs is so complex that a 'traditionally compiled' program just isn't able to take proper advantage of this architecture. It's the software industry's and the processor maker's call to ensure up-to-date and fairly compiled as well as optimized applications that take as much advantage as possible of each processors benefits. FlasK MPEG showed that a frog can suddenly turn into a prince once optimizations were made. It teaches us to take every benchmark result with a grain of salt, because we don't know when or if the underlying software will be enhanced with new processor-optimizations. It might be tomorrow, but it also might be never. So far AMD was suffering from this situation much more than Intel, which is certainly one important reason why AMD will implement SSE2 into the next generation CPU's and not a proprietary own solution. However, it is also important to realize that Athlon does extremely well even when it runs software that was NOT optimized for it!

When you considering the price difference, I am certain that AMD still wins in most situations and the performance difference in this situation is minimal.. imho

[wildone]

Alex99
I am sorry but that was what everyone was fussing about before ,I have yet to see a review where a pentium 4 beat an athon in mpeg encodeing , the only thing I saw it win in was spreadsheets and the such.I also have read somewhere there were talks of a 333 fsb chips or board in the works , or the such , may just be my imagination though , I guess if it was a highly SSE optimized program that the P4 would win but if not the TB still is strong enough for me.
wildone

[Alex99]

Originally posted by wildone
Alex99
I am sorry but that was what everyone was fussing about before ,I have yet to see a review where a pentium 4 beat an athon in mpeg encodeing , the only thing I saw it win in was spreadsheets and the such.

http://www6.tomshardware.com/cpu/01q3/010919/tualatin512-12.html

[wildone]

thank you alex99 ,I really had only saw the high speed of the tb's in the past , it has been a while.

wildone

[wildone]

Wait ,Alex , maybe its not the bandwidth as much as its the pure MHZ , I mean , I bet it would be a whole lot closer if they were set up in a clock to clock fight , I dunno though ,I just love amd ,I know the bandwidth is holding it back some , but do you think the new KT266A chipset will make a difference?The benchmarks I saw amazed me coming from DDR , you can see them here

http://www.lostcircuits.com/motherboard/via_kt266a/4.shtml

Also do you know of any video encodeing benchmarks?

wildone

[Alex99]

It's not that I think the P4 is superior to the Athlon, on the contrary. Also, alot of the benchmarks, has the Athlon 1.4 beating the P4 1.8. But the two particular areas that I care about the most, memory bandwidth and MPEG, the P4 clearly shows its dominance.

[Alex99]

Originally posted by wildone
but do you think the new KT266A chipset will make a difference?

I think the KT266A boost the Athlon memory performance greatly. I am anxiously waiting for it to come out. Hopefully, it'll narrows the gap between Athlon and P4 memory performance.

[UnseenMenace]

Originally posted by Alex99
It's not that I think the P4 is superior to the Athlon, on the contrary. Also, alot of the benchmarks, has the Athlon 1.4 beating the P4 1.8. But the two particular areas that I care about the most, memory bandwidth and MPEG, the P4 clearly shows its dominance.

But the test that you posted which is on Toms Hardware I believe uses the FLASK MPEG4 Encoder which has been optimised by Intel for the Pentium 4 which does not make this a 'fair' comparison as the software which was run during these tests was not optimized for the 'Athlon'.. This if anything clearly shows that the P4 requires software optimized for it to perform well. It is worth noting that Flask has not given Intel permission to release the optimized codec and has no plans itself.

This is a copy of the letter sent to toms hardware from Intel providing the re-written codec (http://www6.tomshardware.com/cpu/00q4/001125/p4-02.html)

The point is that Intel was made supplying Pentium 4 optimized executables which is an unlawful way to show this CPU in a better light and those new results with Intel-optimized executables merely show how much the performance of Pentium 4 can be improved once optimized code is being used.
It is worth remembering that so far AMD has not released Athlon-optimized executables, which is why the latest FlasK-benchmark results are in no way useful to compare the performance of Pentium 4 and Athlon right now.

If the situation with FlasK MPEG should be exemplary for Pentium 4's performance - and why shouldn't it? - then it really counts on the software industry to at least re-compile their applications to make Pentium 4 look a whole lot better. Once that's done Pentium 4 has a good chance to become a success even at its current clock speeds. Of course this is easier said than done. Which software maker would supply its customers with a new and free version of its product, although it only took a re-compilation? Which software maker will even bother to do that for the time being? After all Pentium 4 systems are very expensive and thus not exactly widely spread. We know that Intel has a very forceful way of 'convincing' other industries to follow them. We will see how much power Intel has right now. For the time being Intel is in the same situation AMD used to be with K6-2 and 3D Now! Without proper support of the software industry it will be hard to make a product such as Pentium 4 successful.

[wildone]

Thank you Unseen , look , I went through most of the articles rel;ated to that and the bus speed for the pentium 4 is on 100 mhz ,I thought it was but here is the link to that.

http://www6.tomshardware.com/blurb/00q4/001128/p4-03.html

wildone

[Alex99]

Originally posted by wildone
Thank you Unseen , look , I went through most of the articles rel;ated to that and the bus speed for the pentium 4 is on 100 mhz ,I thought it was but here is the link to that.


Yeah, but it's quad pumped for an effective 400. If you wanna look at it that way then the Athlon is only 100 and 133 too, but it's doubled pumped.

[Alex99]

Originally posted by UnseenMenace


But the test that you posted which is on Toms Hardware I believe uses the FLASK MPEG4 Encoder which has been optimised by Intel for the Pentium 4 which does not make this a 'fair' comparison as the software which was run during these tests was not optimized for the 'Athlon'..

That's a good point and something to take into consideration. I didn't know that Flask was optimized for P4.

[Alex99]

UnseenMenace,
Your remarks are very professional and informative. Do you work in the PR at AMD? :)

[UnseenMenace]

I wish I did, :)

I was also looking to build a system to do encoding recently and saw many reviews which displayed a massive difference in the performance of the Pentium 4 using the Flask mpeg4 codec, so I did some research on the matter and discovered information about the optimized codec.

From what i can understand the Pentium 4 does seem to perform very slightly above the AMD with the standard codec however the difference is not very big at all and certainly not as much as is suggested by Intel. So when considering the price I decided on a high end Athlon with lots of memory and great cards.