Joined
·
8,814 Posts
This time, fast fourier tests by real live scientist dweebs 
Anybody care to wade through it?
http://www.fftw.org/speed/
Anybody care to wade through it?
http://www.fftw.org/speed/
New G5/AMD 64/P4 benches
1 - 20 of 22 Posts
Joined
·
7,114 Posts
If I cross my eyes... I see a sail boat.
Joined
·
8,814 Posts
Those jpegs are a problem?! And, he just grabbed them off the test results. So, you expect him to download them, open them in Graphic Cinvertor, scale them down, post them on some FTP space AND THEN embed them? Come on, people. It's only a forum. It's no big deal.
Anyway, I dont understand what each coloued line means, and I;d wager macDoc doesn;t know all of them. but, here's the Coles Notes version:
The G5 is FASTER. It's coloured lines GO HIGHER.
That's good enough for me
Anyway, I dont understand what each coloued line means, and I;d wager macDoc doesn;t know all of them. but, here's the Coles Notes version:
The G5 is FASTER. It's coloured lines GO HIGHER.
That's good enough for me
Joined
·
160 Posts
Well, you can also scale a JPG or GIF image in HTML.
Say the image is 500x500 pixels, you just add Height="250" Width="250" to the IMG SRC tag. that would scale it in half.
You don't have to open it up in Graphic Converter or Photoshop at all.
But yah, those images mean nothing on their own.
Say the image is 500x500 pixels, you just add Height="250" Width="250" to the IMG SRC tag. that would scale it in half.
You don't have to open it up in Graphic Converter or Photoshop at all.
But yah, those images mean nothing on their own.
Joined
·
6,788 Posts
Those jpegs are a problem?! And, he just grabbed them off the test results. So, you expect him to download them, open them in Graphic Cinvertor, scale them down, post them on some FTP space AND THEN embed them?
If that is what it takes. The entire process you just described would take a minute, maybe two. If he had simply linked to the images instead of posting them, it would have taken even less time.
Come on, people. It's only a forum. It's no big deal.
It is a big deal, because the whole point of a forum is discussion. Throwing the formatting out of whack (causing side-scrolling to be necessary for example) decreases the readability of the thread, and makes discussion less fluid, and (at least partially) defeats the point of the forum.
I dont understand what each coloued line means, and I'd wager Macdoc doesn;t know all of them. but, here's the Coles Notes version: The G5 is FASTER. It's coloured lines GO HIGHER.
The coloured lines in each of those graphs are all from the G5. The only thing the G5 is beating on those two particular graphs, is itself. Macdoc also seems to have picked out two (of not really that many) that show the G5 performing really, really well. On other tests, the other processors coloured lines GO HIGHER and in most of the tests the G5 is higher, but not stupendously so. But again, it is impossible to tell that without having read the article, or having some kind of reference provided.
--PB
If that is what it takes. The entire process you just described would take a minute, maybe two. If he had simply linked to the images instead of posting them, it would have taken even less time.
Come on, people. It's only a forum. It's no big deal.
It is a big deal, because the whole point of a forum is discussion. Throwing the formatting out of whack (causing side-scrolling to be necessary for example) decreases the readability of the thread, and makes discussion less fluid, and (at least partially) defeats the point of the forum.
I dont understand what each coloued line means, and I'd wager Macdoc doesn;t know all of them. but, here's the Coles Notes version: The G5 is FASTER. It's coloured lines GO HIGHER.
The coloured lines in each of those graphs are all from the G5. The only thing the G5 is beating on those two particular graphs, is itself. Macdoc also seems to have picked out two (of not really that many) that show the G5 performing really, really well. On other tests, the other processors coloured lines GO HIGHER and in most of the tests the G5 is higher, but not stupendously so. But again, it is impossible to tell that without having read the article, or having some kind of reference provided.
--PB
Joined
·
8,814 Posts
I understand how the side scrolling thing can be a pain on like, a G3 iMac or something. But, the pics are not THAT big. And, it wouldn't take "a minute"; it would take TWO!
Yes, each chart represents a different computer. Each colour on the chart represents a particular test . What each test IS is the part that I don't understand. But, the G5 seems to beat 'em.
Class dismissed.
Yes, each chart represents a different computer. Each colour on the chart represents a particular test . What each test IS is the part that I don't understand. But, the G5 seems to beat 'em.
Class dismissed.
Joined
·
31,035 Posts
Fast Fourier Transforms are good measures of intensive math operations and the plots are various transforms they tested on. FFTs are used in things like loudspeaker crossover circuit design amongst many other complex systems like signal processing.
I picked the two appropriate images being the Apple G5 dual and the equivalent PC out of the dozens on the link.
What's so hard to understand about that.
![]()
Thie image takes up about 1/12th of one of my screens and the text along the edge is easy to read.
If you want it scaled, scale it and post it and I'll strip these otherwise this is an easy visual indication of the kind of performance that the cluster system exhibited.
The article is there, you can get into the details of areas where one or the other processor has some advantages in a particular math function if you really want to geek it..
Getting apps to take full advantage of this performance is another thing altogether but at least the fundamental performance is there to be used.
I would like to see the impact of caches here - the same processor but with a different cache structure.
I notice that a couple of the tests show Apple's concentration on making use of caching. When they couldn't clock they cached -
[ November 12, 2003, 09:18 PM: Message edited by: macdoc ]
I picked the two appropriate images being the Apple G5 dual and the equivalent PC out of the dozens on the link.
What's so hard to understand about that.
Thie image takes up about 1/12th of one of my screens and the text along the edge is easy to read.
If you want it scaled, scale it and post it and I'll strip these otherwise this is an easy visual indication of the kind of performance that the cluster system exhibited.
The article is there, you can get into the details of areas where one or the other processor has some advantages in a particular math function if you really want to geek it..
Getting apps to take full advantage of this performance is another thing altogether but at least the fundamental performance is there to be used.
I would like to see the impact of caches here - the same processor but with a different cache structure.
I notice that a couple of the tests show Apple's concentration on making use of caching. When they couldn't clock they cached -
[ November 12, 2003, 09:18 PM: Message edited by: macdoc ]
Joined
·
17,353 Posts
Everybodies shirt tails inna knot??
12" iBook, and I see all just fine. Whadda ya onna PDA or something?
Interesting, I'd be interested in hearing some more "deciphering" from ones who know.
12" iBook, and I see all just fine. Whadda ya onna PDA or something?
Interesting, I'd be interested in hearing some more "deciphering" from ones who know.
Joined
·
31,035 Posts
Yeah I tried it in Explorer as well as Safari and the image is just a tad past the edge as it is on the original page
![]()
Seeing the two graphs together provides a good visual comparison.
Instead of worrying about the image perhaps someone could explain why Photoshop which SHOULD make good use of the top notch math capability on the G5 .....isn't.
Seeing the two graphs together provides a good visual comparison.
Instead of worrying about the image perhaps someone could explain why Photoshop which SHOULD make good use of the top notch math capability on the G5 .....isn't.
Joined
·
3,364 Posts
i don't know what all the funny colored little lines are -- i guess the g5 is slightly faster in some area -- but then it looks like the pc is faster in other 'areas'. this is complete gibberish.
gord guide?
gord guide?
Joined
·
2,126 Posts
i love peewee's playhouse!
and i love those graphs!
and i love those graphs!
Joined
·
6,788 Posts
Side scrolling is annoying on any sized screen. You seem to be missing my basic point though.
Poor formatting = ugly page = not as easy to read = annoying as hell = a hinderance to discussion.
Fast Fourier Transforms are good measures of intensive math operations and the plots are various transforms they tested on. FFTs are used in things like loudspeaker crossover circuit design amongst many other complex systems like signal processing.
I picked the two appropriate images being the Apple G5 dual and the equivalent PC out of the dozens on the link.
What's so hard to understand about that.
Nothing at all. But it is interesting that you only posted graphs of one test out of 24, on which the G5 did really, really well, and presented it as if it is the only test that matters.
The G5 does well on fast fourier transforms, but it is also supposed to do well on Floating Point and Integer Performance operations, and an Itanium 2 can whup a G5 the former[1], and a P4 can whup a G5 on the latter[2]. My point here (again) is that these graphs are completely meaningless unless you read the article. For the record, the G5 is ahead in a lot of the tests, but it is only really ahead in about 4 of the 24.
--PB
[1]SPECfp_base2000 tests
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 1444 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1267 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerPC G5 (Single 2.0GHz) - 840 (apple.com/powermac)
[2]SPECint_base2000 test
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 824 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1205 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerMac G5 (Single 2.0GHz PowerPC G5) - 800 (apple.com/powermac/)
Poor formatting = ugly page = not as easy to read = annoying as hell = a hinderance to discussion.
Fast Fourier Transforms are good measures of intensive math operations and the plots are various transforms they tested on. FFTs are used in things like loudspeaker crossover circuit design amongst many other complex systems like signal processing.
I picked the two appropriate images being the Apple G5 dual and the equivalent PC out of the dozens on the link.
What's so hard to understand about that.
Nothing at all. But it is interesting that you only posted graphs of one test out of 24, on which the G5 did really, really well, and presented it as if it is the only test that matters.
The G5 does well on fast fourier transforms, but it is also supposed to do well on Floating Point and Integer Performance operations, and an Itanium 2 can whup a G5 the former[1], and a P4 can whup a G5 on the latter[2]. My point here (again) is that these graphs are completely meaningless unless you read the article. For the record, the G5 is ahead in a lot of the tests, but it is only really ahead in about 4 of the 24.
--PB
[1]SPECfp_base2000 tests
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 1444 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1267 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerPC G5 (Single 2.0GHz) - 840 (apple.com/powermac)
[2]SPECint_base2000 test
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 824 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1205 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerMac G5 (Single 2.0GHz PowerPC G5) - 800 (apple.com/powermac/)
Joined
·
31,035 Posts
???
[1]SPECfp_base2000 tests
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 1444 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1267 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerPC G5 (Single 2.0GHz) - 840 (apple.com/powermac)
[2]SPECint_base2000 test
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 824 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1205 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerMac G5 (Single 2.0GHz PowerPC G5) - 800 (apple.com/powermac/)
•••••••
It's in exactly the same scale as the original page which is very readable
One is labelled with one machine - the other is labelled with the other ...formatting???
Take the page authors to task.
or
clean it up, scale it, post it and show the graphs you think are more balanced.
[1]SPECfp_base2000 tests
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 1444 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1267 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerPC G5 (Single 2.0GHz) - 840 (apple.com/powermac)
[2]SPECint_base2000 test
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 824 (http://www.spec.org/cpu2000/results/res2003q3/)
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1205 (http://www.spec.org/cpu2000/results/res2003q3/)
Apple PowerMac G5 (Single 2.0GHz PowerPC G5) - 800 (apple.com/powermac/)
•••••••
It's in exactly the same scale as the original page which is very readable
One is labelled with one machine - the other is labelled with the other ...formatting???
Take the page authors to task.
or
clean it up, scale it, post it and show the graphs you think are more balanced.
Joined
·
6,788 Posts
[1]SPECfp_base2000 tests
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 1444
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1267
Apple PowerPC G5 (Single 2.0GHz) - 840
[2]SPECint_base2000 test
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 824
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1205
Apple PowerMac G5 (Single 2.0GHz PowerPC G5) - 800
Since this data is from different sources (apple didn't submit their results to SPEC), there was no graph.
In a nutshell, on the SPECfp_base2000 test which measures floating point operations, an Itanium 2 running at 1.4Ghz scored a 1444, while a single G5 running at 2.0 Ghz scored an 840.
On the SPECint_base2000 test, which measures Integer performance, the Itanium 2 only scored an 824, but the Single G5 was still lower with an 800, and the P4 trounced them both with a score of 1205.
Higher numbers, as previously established, go faster.
Sorry, I guess I should have made that more clear.
--PB
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 1444
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1267
Apple PowerPC G5 (Single 2.0GHz) - 840
[2]SPECint_base2000 test
Dell PowerEdge 3250 (1.4GHz/1.5MB, Itanium2) - 824
Dell Precision Workstation 360 (3.2 GHz P4, DDR400) - 1205
Apple PowerMac G5 (Single 2.0GHz PowerPC G5) - 800
Since this data is from different sources (apple didn't submit their results to SPEC), there was no graph.
In a nutshell, on the SPECfp_base2000 test which measures floating point operations, an Itanium 2 running at 1.4Ghz scored a 1444, while a single G5 running at 2.0 Ghz scored an 840.
On the SPECint_base2000 test, which measures Integer performance, the Itanium 2 only scored an 824, but the Single G5 was still lower with an 800, and the P4 trounced them both with a score of 1205.
Higher numbers, as previously established, go faster.
Sorry, I guess I should have made that more clear.
--PB
1 - 20 of 22 Posts
Join the discussion
