20 seconds broke on Super PI

[Kunaak]

20 seconds broke on Super PI, now thats a hell of a landmark



http://www.xtremesystems.org/forums/...217#post856217

[RaptorRaider]

Can't wait until Intel releases Pentium M's desktop brothers; with performance like this they'll reconquer the enthusiast market fast.

[saaya]

damn thats nice!!!! congrats!

this was done by newbeatle right?

[joezxk]

wa kaka

that is a revelotion!!!

it's under 200'c?

[austinbmxnig]

i wonder if that would be good in 3d at all

[(sin)morpheus]

Wow, amazing time! I think it would do quite well in gaming. Go short pipes!

[$a1Ty]

AMD fan boy here, intel has really done themselves harm, if only they released dual core dothan's now instead of the joke that is smithfield, dothan is awesome,

*eagerly waits for yonah*

yes and rock on effiecient short pipes, none of this long crap, its not the size that counts

[Nubius]

Go short pipes!

??? I thought intel had the longer fatter pipes and AMD had the shorter thinner pipes........

[MusicIsMyLife]

awesome

[Flib]

??? I thought intel had the longer fatter pipes and AMD had the shorter thinner pipes........

yes intel has long pipes but with the pentium m they go the short pipe way!

[Leverpospade]

Nice!!!!!! Congratulations with the nice work - keep up the good work

[Nubius]

yes intel has long pipes but with the pentium m they go the short pipe way!

Ahh I see thanks for the insight......so basically they learned a thing or two from AMD? I kid I kid don't get up in arms about it :P

[robberbaron]

It's too bad SuperPI is a broken benchmark and that Pentium4's would beat the PM with ease if it truly tested the FPU. I'd like to find things that the PM is good at, other than SuperPI. It seems like a one-trick pony to me, though gaming becnhmarks are pretty impressive.

Still pretty interesting though.

[Xassius]

Ahh I see thanks for the insight......so basically they learned a thing or two from AMD? I kid I kid don't get up in arms about it :P

Not really. This PM architecture was based in the Pentium 3. Then the pipes just got longer from there............................................. .................to 31 pipes on the prescott.

[teqguy]

As far as pipeline width and length is concerned, there really is no superiority.

Shorter, thinner pipelines are more adequate for latency intensive tasks(which is why the Athlon typically benchmarks better in games), while longer, wider pipelines are better for data hauling tasks, such as rendering and encoding.

Outside of benchmarks, though, longer and wider pipelines are a lot more practical, because real world usage will never need to be as latency intensive as the processor is capable of... unless of course we find a way to percieve motion at framerates excess of 60FPS.

[saaya]

Ahh I see thanks for the insight......so basically they learned a thing or two from AMD? I kid I kid don't get up in arms about it :P

lol, no, the dothan pentium m is based on the banias pentium m wich was an improved pentium 3 tualatin.

netburst wasnt really a mistake... intel just seems to have hit the limit of the architecture now and it seems they arent able to keep netburst scaling higher and keeping the power consumption low at the same time, so they go a step backwards to the p6 architecture (pentium 3) ,turn a tad to the right and make two steps forward wich gets them further to where they want to go than netburst would have gotten them

i think the problem with netburst was that they had to lengthen the pipeline to reach higher clocks, but the longer your pipeline gets the more complicated it gets to keep the pipeline full, plus higher clocks are hard to reach without losing too much energy to leakage...

i think its basically a result of todays manufacturing processes, you cant reach the speeds with current processes that a netburst processor would need to perform fast enough to be able to compete, without reaching insane leakege levels...

since the manufacturing process is limiting them to a certain speed they are now trying to get as much performence as possible at lower clockspeeds wich means a shorter pipe

if the current processes wouldnt have such a high leakage with the netburst architecture, wich intel didnt expect, then prescott would run a lot cooler and clock a lot higher, at least 30% for each i think.


back on topic though, amazing results, and while the pentium m seems to be weak in some tasks, its easy to see that it has a hole lot of potential wich is why intel chose it as the base for their new architecture after netburst

[Cooper]

I'd like to find things that the PM is good at, other than SuperPI

Some tests compared to P4 and AMD K8:http://www.gamepc.com/labs/view_cont...70ct479&page=1

[$a1Ty]

Outside of benchmarks, though, longer and wider pipelines are a lot more practical, because real world usage will never need to be as latency intensive as the processor is capable of... unless of course we find a way to percieve motion at framerates excess of 60FPS.

for starters human eyes can detect about 72fps

secondly there is such a thing as having too long a pipeline, the prescott is rediculous

[hawtrawkr]

i dont even know why i still look at these threads....... they just make my wallet hurt.

[teqguy]

for starters human eyes can detect about 72fps

secondly there is such a thing as having too long a pipeline, the prescott is rediculous

Um, no...

A human's visual perception can be tricked into percieving motion at as little as 24FPS, which is why video is set at a comfortable 29.7FPS.

If you were able to visualize 72 frames per second, the concept of motion blur would not exist to you.


As far as the Prescott's pipeline length being rediulous, what's rediculous is the fact that you don't realize that the pipeline was also widened. They could've easily just lengthened the pipeline so that the frequency would scale higher.

Despite that, as I said previously, the only benefit of a shorter pipeline is apparent in latency intensive tasks, which are non-existent outside of benchmarks.

[PowerSlide]

awesome

4ghz possible?

[teqguy]

I doubt it, considering the stock clock is 1.8Ghz.

Besides, what's another 400Mhz? Practically nothing.

[saaya]

Um, no...

A human's visual perception can be tricked into percieving motion at as little as 24FPS, which is why video is set at a comfortable 29.7FPS.

If you were able to visualize 72 frames per second, the concept of motion blur would not exist to you.


As far as the Prescott's pipeline length being rediulous, what's rediculous is the fact that you don't realize that the pipeline was also widened. They could've easily just lengthened the pipeline so that the frequency would scale higher.

Despite that, as I said previously, the only benefit of a shorter pipeline is apparent in latency intensive tasks, which are non-existent outside of benchmarks.

hmmm then why do cpus with a shorter pipeline tend to perform better in games and other apps? i thougt its because games love low latencies.

[White]

Super! Fantastic perfomance!

[saaya]

hey white, welcome to XtremeSystems