Athlon FlasK optimisations
When I looked here about 10 hours ago, someone has posted a link to the page on amdzone.com which talked about the optimisations those guys are doing to make FlasK run faster on an Athlon. Now that message seems to have disappeared. Was it removed deliberately?
I've found the page. It's at:
I've got no idea what happened to the link to AMDZone. I am actually not quite as watchful as you guys think. Wish I had the time.
What is more important is the fact that I have tried out the latest (as of 12:35 AM) Athlon-optimized version done by Alex from AMDZone.
Unfortunately Athlon is still not touching the P4-results. The Athlon optimized x87-code on Athlon 1200/133 DDR is a tiny bit slower than the P4-optimized x87 code on P4 1500. However, the 3DNow! optimized Athlon code on Athlon 1200/133 DDR is way slower than the double precision SSE2-optimized code on P4 1500.
So far about that.
AMDZone is working on those optimizations for more than 24 h now. Intel had only one night. I guess it seems as if AMD lost this round.
I am still hoping to receive a different AMD-optimized version of FLasK directly from AMD tomorrow.
your words really disappointed me. did you even forget that there is a 300mhz gap between P4 1.5ghz and Athlon 1.2ghz ddr? take this into consideration, please.
maybe the Athlon 1.2ghz ddr is so good that we begin to see it as a fair counterpart of the 1.5ghz P4. however, a reviewer should always be aware of this *FACT*.
even your own non-SSE2-optimized test using Intel-recompiled FlasK actually showed that Athlon 1.2ghz ddr is not the loser at all, if the 3000mhz gap is taken into consideration.
Flask /DivX MPEG4 Video Compression
High Quality iDCT x87 (fps)
_______________Re-Compiled___Performance Gap___Clock Gap
(Data source: www.tomshardware.com)
Some are ignorantly happy,
While some, happily ignorant.
You guys just don't WANT to believe that the P4 is NOT a complete piece of crap. No ladies, the P4 beat the Athlon and that's it! AMD is not able to deliver it at higher clock speeds right now and the high clock of P4 is an integral part of its design. Just accept it. Athlon has lost this round and you can twist it and turn it as much as you want.
We are not fighting BELIEFS or RELIGIONS here, we are simply comparing benchmark results. Your comment shows that you have lost your objectivity. If the Athlon would have beat the P4 it would have been great, but if it loses then it's still great, because after all there's the 300 MHz gap?
This reminds me of the Black Knight in Monty Python's Holy Grail.
<P ID="edit"><FONT SIZE=-1><EM>Edited by Tom on 11/28/00 10:22 AM.</EM></FONT></P>
Heh heh. It's only a flesh wound.
Thank you Dr. Thomas Pabst. I'm glad you're finally standing up for the P4 when the proof is right there that when using software optimized for SSE2, the P4 rocks.
I think Intel was a little nuts to not have as much backwards support as they do making non-optimized code pretty poorly supported. But then, in a year from now, who will care about non-optimized code? Especially since even just fundamentally SSE2 will be so easy to implement.
I like AMD as much as I like Intel. AMD seemed to have struck a good blow with their Athlons and Durons. But now that the gloves are off and Intel has shaken a bit of the concussion, I think we're all going to see just what really happens when someone tries to make Intel look bad.
Intel dropped their broken sword labeled 'P3' and whipped out a new one labeled 'P4'. And unless AMD has something up their sleeve, it's only a matter of time before it's all over and Intel is once again not only 'King Of The Hill', but also the only one really even on the hill.
It's too bad AMD and Intel can't just join forces, produce one heck of a chip, and force the Macintosh to cease existing.
I wonder if for X-Mas, Mac will dare claim that they're faster than a Pentium4. I still can't believe that they claimed to be faster than a PC ... ever.
For that matter ... Tom, can you give us a comparison of a Mac vs. both AMD and Intel PCs running similar software? I know it'd be hard to find a fair comparison, but it would be interesting to see.
- Anything can be fixed with duct tape, a swiss army knife, and WD-40.
Heh heh. I'm surprised that I didn't get flamed to Hellen Bach repeatedly from every possible direction on that one. Am I that lucky? Or has every AMD and Mac fan hung their heads and accepted defeat?
Oh, and I was at least very close to right about Macs actually daring to lie about their supremacy, wasn't I? Heh heh.
- Sanity is purely based on point-of-view.
The sad part is that I don't even hold anything against AMD. If they produced their own chipsets (preferably that support multiple-processor motherboards) then I'd probably look at putting together an Athlon system.
But the simple fact is that with one quick recompile to include SSE2 support and P4 optimizations, the P4 totally reemed the Athlon. Yeah, so the P4 is 300MHz faster. I don't even see AMD making a chip that fast. And that's half the point.
Intel released a chip that was by far a faster clock than AMD's. And AMD has no counter-chip to one-up Intel. Their only hope was to prove that the Intel chip sucked because of it's crappy floating-point processing. But I think this Flask optimization battle has proven that the P4 is far from the piece of junk AMD would like people to believe.
And with just a few hours spent optimizing the software, the P4 looked great. It took AMD fans a lot longer than that to optimize their software and they still can't beat Intel's quickly done optimizations.
Imagine if the Intel optimized version had as many man-hours put into it as the AMD one did! Heh heh. It's a good thing we haven't seen the benchmarks from something like that, or there'd be AMD fans committing suicide!
The simple fact is: Intel makes a more powerful chip than AMD when optimized software is used.
Hmm ... it makes me wonder if someone has made a disassembler that will then reassemble with P4 optimizations. Heh heh. That'd be a nifty trick to just run your program through that strainer and violla: A P4 optimized executable.
- Sanity is purely based on point-of-view.