Int crown move to IPF with a 1300/2100

[juin]

As i say ealier this new itanium is fast really fast in FPU performance is unrival about 40% that anyone. The 1 to 5 are all Itanium except nb 2 who is a Power4+ 1.7GHZ.

There few intersesting part 1 opteron is part of the VS itanium as the box will be sold for 5000$.2 P4 are test with canterwood but with DDR 2100 as most workstation dont use DDR400 personaly i think under 4GB DDR333 can be use now.Page 12 benchmak is nice also they show a 3 times faster Itanium but just beat P4 3.06 GHZ.Most of workstation apps wont be much faster on Itanium that it was on a P4.We will have to wait a others year or so before see itanium lead in most of the benchmark in all market

http://www.hp.com/products1/itanium/pdfs/infolibrary/itanium_workstations_performance_june_25_2003_FINAL.pdf

http://www.theinquirer.net/?article=10250


[-peep-] french

 

[eden]

I'm not impressed.
Now you have better cache, and 50% more clock speed and all you get is a scalar result of 50% more FPU points! I expected MORE.

Upon second inspection, the INT performance did go higher than 50%, which, I guess, is very nice.
--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue: <P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 06/30/03 11:20 AM.</EM></FONT></P>

 

[Mephistopheles]

Is that performance scale truly linear in nature? (I mean, SPEC?)

Besides, I think you're missing the point. The point is that McKinley's FP score was already impressive, and +50% on that is a staggering 2100 or so points, and that's a new record.

 

[juin]

according to Hp and intel they will be a increase by 30% to 50% over 1GHZ mackinley

[-peep-] french

 

[juin]

On DIVX benchmark look a Pentium 100 vs P4 3066 mghz there a 30X difference equal to the frequency disparity even if thre much more units on P4 more cache and much faster I/O no CPu scale linear for a long time on a platform.Consider also that they still use the same bandwith

[-peep-] french

 

[Mephistopheles]

That makes sense! The main core clock increase isn't everything. 3.0Ghz P4 with 30x the performance of a P100 is a good indication of that... The 3Ghz has more of everything, not just of clock, to really get that 30x increase.

 

[juin]

but on IPF only the clock speed and L3 cache have increase by 50% and real performance increase by 50% to 30% on X86 we need the more bandwith more units better chipset and many others improvement.Unlike X86 it take much less the give a good scaling of the chip.

[-peep-] french

 

[eden]

It seemed to me that the Pentium had near the IPC of a Pentium 4 if I am not mistaken. Also take SSE2 into account, as other benches, demonstrate P4 3GHZ being over 100 times faster.
What about the Athlon XP 2GHZ compared to a 100MHZ P1?
--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue: <P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 06/30/03 04:25 PM.</EM></FONT></P>

 

[eden]

It went up a expected but the other IPC improvements would've done more.
However now that I thought about it, yes the lack of a new memory platform and bus does make a difference, and if it did get an upgrade on that, it'd likely get more than 50% easily.

--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue:

 

[juin]

In theory it have 100 times more power toms made benchmark it self look at the score on Divx

[-peep-] french

 

[eden]

There is no DivX score, IIRC.

--
I am my own competition. -VJK