06-Dec-2016 21:46 GMT.
UNDER CONSTRUCTION
Anonymous, there are 177 items in your selection [1 - 50] [51 - 100] [101 - 150] [151 - 177]
[Forum] Amigaone v Peg BenchmarkANN.lu
Posted on 08-Mar-2003 00:51 GMT by Skyraker177 comments
View flat
View list
Peg and a1 have been benchmarked.... read on and make your own conclusions http://personal.inet.fi/cool/pekosbil/a1benchmarks.htm
Amigaone v Peg Benchmark : Comment 1 of 177ANN.lu
Posted by Alkis Tsapanidis (212.205.250.214) on 07-Mar-2003 23:58 GMT
Where are the BYTEmark benchmarks for the Peg? I can say they are QUITE interesting.
Amigaone v Peg Benchmark : Comment 2 of 177ANN.lu
Posted by Anonymous (62.42.228.20) on 08-Mar-2003 00:24 GMT
The dnet client performs exactly the same on two 600MHz PPC 750CEx CPUs. Truly amazing!

What was the point of this again?

BTW, not that it'll make any significant difference, but was the Pegasos running dnetc on MorphOS?
Amigaone v Peg Benchmark : Comment 3 of 177ANN.lu
Posted by Peter Gordon (217.158.203.218) on 08-Mar-2003 00:41 GMT
Gosh, look. Identical performance from the same processor with similar chipsets. And look! The G4 XE 800 is faster than both G3 600s.

What does this prove? Nothing at all. Benchmarks are pointless. Both MorphOS and AmigaOS4 are going to feel very fast on their respective machines.

No doubt these tests will start a flame war anyway...
Amigaone v Peg Benchmark : Comment 4 of 177ANN.lu
Posted by Sam Smith (62.64.222.247) on 08-Mar-2003 00:52 GMT
In reply to Comment 2 (Anonymous):
There have been some comments that the G3 in the Pegasos is faster than the G4 based Amiga One. These tests go someway to disproving that.

This particular 'news' item came about at exactly the same time that it was announced that the Pegasos could only support a G3 in its current configuration.

---
Sam
Amigaone v Peg Benchmark : Comment 5 of 177ANN.lu
Posted by Anonymous (151.203.121.39) on 08-Mar-2003 00:52 GMT
In reply to Comment 3 (Peter Gordon):
proves nothing exactly, so find out why Buck is insisting that his G3600 is faster than Eyetech's G4800
Amigaone v Peg Benchmark : Comment 6 of 177ANN.lu
Posted by Hagge (213.112.60.58) on 08-Mar-2003 01:02 GMT
Actually i don't see what this is supposed to show us either. The first test is only on amigaone, and the dnetc doesn't mention which client is used and under what os, and if that is a program there different mb design should make a difference or not.

Nothing to see here, move along.
Amigaone v Peg Benchmark : Comment 7 of 177ANN.lu
Posted by Situation (151.38.207.240) on 08-Mar-2003 01:10 GMT
Situation!
Amigaone v Peg Benchmark : Comment 8 of 177ANN.lu
Posted by Megol (130.240.196.54) on 08-Mar-2003 01:16 GMT
In reply to Comment 3 (Peter Gordon):
Yes most benchmarks are pointless and these especially so :-)
BYTEmark: If there is an example of an outdated benchmark this is it...
RC5-72: Perhaps interresting if one only use a computer for encryption/decryption... This does not map to anything a ordinary user do with their computer. However I think the G4 800MHz scores are really low?
OGR: Really...

I can not understand that any AmigaOne/Pegasos buyer would be interrested in benchmarks, performance should logically not be of any importance as even a low end x86 machine is faster and _much_ cheaper. For the same money one can buy a AmigaOne G4 XE one can get a dual AMD xp2400+ with 1GByte RAM and a 120GB hdd!
Amigaone v Peg Benchmark : Comment 9 of 177ANN.lu
Posted by Anonymous (62.42.228.20) on 08-Mar-2003 01:49 GMT
In reply to Comment 4 (Sam Smith):
> There have been some comments that the G3 in the Pegasos is faster than the G4
> based Amiga One.

I have not seen any such nonsense comments about CPUs (G3 (which "G3"?) being faster than G4 (which "G4"?)).

I have seen comments about the current Pegasos being an altogether faster system than the currently available Teron boards, which, until the Teron boards are released with revised ArticiaS, most likely is true. What numbers we're talking about, and if they're significant is anyone's guess though. What I'm more concerned with is data integrity in systems based on the "old" ArticiaS, like every Teron board hitherto sold by Eyetech. THAT would be an interesting and VERY important test.

A dnetc run is absolutely meaningless to compare systems with, as it's almost 100% CPU dependent.

> These tests go someway to disproving that.

Nonsense. It's just a dnetc run.
I.e. "how fast does this CPU and dnet client crunch RC-5 keys and calculate optimal goulomb rulers?".

> This particular 'news' item came about at exactly the same time that it was
> announced that the Pegasos could only support a G3 in its current
> configuration.

Nonsense again. If you have followed the news the last couple of weeks, you'd have seen that the Pegasos I will have a G4 upgrade module (although if the Peg II upgrade will be $200 *including*a G4, I personally don't expect to see much demand for that). No NEW Pegasos I boards will be SHIPPED with a G4 module though, because the Pegasos I is no longer made (duh!).


What's all this fighting over a couple of very similar motherboards all about anyway? Is it just that one dealer has decided to rename one of the boards to something with "Amiga" in it, and create a monopoly in the process, that makes people go so awfully defensive on eachother? Some of you guys sometimes act as if there are still "Amigas" around to "defend" and you're 12 years old for crying out loud!

The more hardware the merrier.
Amigaone v Peg Benchmark : Comment 10 of 177ANN.lu
Posted by Christophe Decanini (24.128.31.27) on 08-Mar-2003 02:48 GMT
This is pointless to rely on these benchmarks to prove anything about BBRV claims:

- The Pegasos used to do these benchmarks does not include the April2 which is supposed to improve northbridge performances.
- The benchmark figuring the Pegasos are heavely depending on the CPU and not on the northbidge performance. Memory tests would be interesting, but once again on a Pegasos with April2. Hard disk and GFX benchmarks would be interesting to do too.

Remember that the overall performance of a system is not limited to the CPU.
We should know that with several examples of crippled memory designs in Amiga computers such as in the A4000.
Amigaone v Peg Benchmark : Comment 11 of 177ANN.lu
Posted by Nate Downes (172.137.210.89) on 08-Mar-2003 03:30 GMT
In reply to Comment 9 (Anonymous):
I would point out that the referred to text was also discussing the April 2'd boards. As none have been disseminated, these benchmarks only display similar setups doing similar jobs.

The slight degredation on the last test could have even been explained away by poor ventilation, it was that slight. A better examination of the benchmarks, in lab conditions, would show more than this.

However, I would note from these that the G4 is not performing as well as it should be, using it's performance on the Macintosh as a reference.
Amigaone v Peg Benchmark : Comment 12 of 177ANN.lu
Posted by ehaines (63.159.189.91) on 08-Mar-2003 03:32 GMT
In reply to Comment 9 (Anonymous):
Somehow I doubt there will be a G4 upgrade module for the Pegasos I...
I mean, really, are they going to go through all that trouble and
expense for probably a few dozen sales? More power to 'em if they do,
but it makes no business sense whatsoever so don't expect it.

Anyway, I've been running my AmigaOne for a couple of months now and
there are no data integrity issues that I've been able to find. Not that
I copy gigabytes of data back and forth every day, but for normal use
nothing bad has happened and that's with DMA enabled.

Speaking of which, I find the speed more than satisfactory...for example,
with MOL, MacOS9 boots in maybe 10 seconds. That's like 10 times faster
than on my real 300MHz G3 PowerMac, maybe more.
Amigaone v Peg Benchmark : Comment 13 of 177ANN.lu
Posted by Senex (62.226.104.158) on 08-Mar-2003 04:32 GMT
In reply to Comment 12 (ehaines):
You forget the possible licensees for the Peg-I-design. For the few hundred Peg-I-users, indeed, such a module wouldn't make sense.
Amigaone v Peg Benchmark : Comment 14 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 04:50 GMT
I wish people would stop using dnetc as a benchmark. Specialized programs like that make horrible benchmarks.

We also don't have any specs on these machines. RAM? Disk? Anything? This information is less than useless.

Does Bytemark measure things like disk and vram i/o?
Amigaone v Peg Benchmark : Comment 15 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 04:53 GMT
BTW just to drive the point home regarding dnetc...

I have seen the dnetc PowerPC code. It's a very tight loop which doesn't stress anything other than CPU. In other words it isn't restricted by memory bandwidth or anything else. If there were memory bus or peripheral bus bandwidth issues on the A1, dnetc would NOT detect it.

Please people, don't use dnetc as a benchmark. You look very foolish doing so.
Amigaone v Peg Benchmark : Comment 16 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 04:57 GMT
In reply to Comment 12 (ehaines):
If the dual G4 upgrade boards were sold before the PII I wouldn't be surprised if they sold 200 of them.

Not to say that would make it worth their while...
Amigaone v Peg Benchmark : Comment 17 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 05:09 GMT
The Apple PowerBook I bought last year gets 11,222,655 nodes/sec OGR and 9,749,662 keys/sec RC5-72

Now you have another reason to stop using dnetc as a benchmark.
Amigaone v Peg Benchmark : Comment 18 of 177ANN.lu
Posted by Anonymous (217.187.194.113) on 08-Mar-2003 06:16 GMT
In reply to Comment 11 (Nate Downes):
>However, I would note from these that the G4 is not performing as well as it>should be, using it's performance on the Macintosh as a reference. It's funny to see G3-700 beats G4-800 in 4 out of 10 Bytemark/nbench tests.I assume a G3 will beat the G4 AmigaOne in at least 7 out of 10 Tests at 800MHz. We need more and different benchmarks to differentiate the specific strengthsand weaknesses. I'm looking forward to the AOne/750FX and Pegasos/April2 results.
Amigaone v Peg Benchmark : Comment 19 of 177ANN.lu
Posted by Remco Komduur (195.241.12.96) on 08-Mar-2003 07:05 GMT
Two things:

1) Benchmarks don't tell anything!! These are ideal test which prove nothing.
2) Similar processor with similar chipset WILL all be as fast.

In the PC world it's quite common to use 3DMarkxxx to benchmark video drivers and they prove even less. Only the stupidity of the average PC-owner and the companies.

Fact: NVidia recently said that severall tests of the newest 3DMark were crap and NVidia had to write code in the drivers to iron out these flaws. It was making the overall driver slower and buggier!!!! There's no doubt in my mind that Ati do the same.
This is a very unhealty situation. Driver performance and programming is dictated with 1 stupid benchmarking program. And whats even worse is that they do this because these headless PC owners follow these crap benchmarks!!

I hope we don't get such a crap situation here.
Amigaone v Peg Benchmark : Comment 20 of 177ANN.lu
Posted by Pekka Nissinen (194.252.183.62) on 08-Mar-2003 07:36 GMT
In reply to Comment 17 (strobe):
Do you have a G4 Mac? The Mac dnetc client is the only one (at the moment) that supports Altivec properly. Even the current WarpUp clients ppc-optimisations were done by some Mac people...
Amigaone v Peg Benchmark : Comment 21 of 177ANN.lu
Posted by Pekka Nissinen (194.252.183.62) on 08-Mar-2003 07:43 GMT
In reply to Comment 2 (Anonymous):
No, they were done under Linux, as it's said on the page. All PPC tests (A1 and Pegasos) were done with exactly the same exe... The only difference is that A1 tests were done under Debian and Pegasos tests under SuSE.
Amigaone v Peg Benchmark : Comment 22 of 177ANN.lu
Posted by catohagen (81.29.32.150) on 08-Mar-2003 08:33 GMT
In reply to Comment 5 (Anonymous):
>so find out why Buck is insisting that his G3600 is faster than Eyetech's G4800

maybe its faster in freefall from a cliff or something..:)

its funny, when some tests are provided and the pegasos doesn't perform as
hoped, current/future pegasos users(see, i didnt use the 'camp' word) are quick to say like

'proves nothing'
'absolutely meaningless'
'pointless to rely on these benchmarks'
'this information is less than useless.'

when BBRV did propose a G3 vs G4 speed competition, current/future A1 users used similar words, wondering if BBRV somehow was right..

Lets stop with this arguing, and do some new tests when OS4 and MOS are
out of beta, and released to end users.. :)
Amigaone v Peg Benchmark : Comment 23 of 177ANN.lu
Posted by Don Cox (62.254.64.5) on 08-Mar-2003 08:47 GMT
In reply to Comment 22 (catohagen):
"Lets stop with this arguing, and do some new tests when OS4 and MOS are
out of beta, and released to end users.. :)"

Something like applying blur to a big image in Image Engineer would be a suitable test. That would test memory and emulator speed as well as CPU speed.
Amigaone v Peg Benchmark : Comment 24 of 177ANN.lu
Posted by DaveP (195.92.67.75) on 08-Mar-2003 08:50 GMT
In reply to Comment 23 (Don Cox):
But not disk access, network connection speed etc.
Amigaone v Peg Benchmark : Comment 25 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 09:02 GMT
In reply to Comment 24 (DaveP):
Well, we're only STARTING to form the page. Come on, give us some slack here.
I'm the fool running the SE at 700MHz, btw.

Instead of everyone yelling that the benchmarks we ARE using are not to be relied on, why don't you come up with some FEASIBLE examples. Something all 4 computers we're testing can run (Mine is overclocked, but with 100MHz memory bus, Pekka is overclocking with 133MHz memory bus, Massimiliano is running at normal frequency with 133MHz memory bus, Gunne is running with 133MHz memory bus (I think so, at least). So of course I will loose out on a memory intensive benchmark (like for instance LU decomposition, see?). On the other hand, I'll check and see if the Pegasos LU decomposition really is as low as the figure reported, or if it's a typo. That would be BAD for any further memory intensive Peggy-testing.

So, to summarize the summary of the summary: Give us some input as to what is acceptable Linux-runnable benchmarks. Compiling is DIFFICULT to test, since ALL the BIG projects will enable different options on different machines depending on a LOT of variables (G4 present, AMD present (don't forget this machine either...), G3 present, what other system libraries and which version, version of compiler, optimization of compiler when it was compiled etc.) So it must be a self-contained package we can all run more or less out of the box.

AmigaOS programs need not apply, because we could only run those in UAE without JIT on 1 out of 5 machines at the moment...
Amigaone v Peg Benchmark : Comment 26 of 177ANN.lu
Posted by DaveP (195.92.67.75) on 08-Mar-2003 09:05 GMT
In reply to Comment 25 (Ole-Egil):
Hey hey hey :-) I was just pointing out that a "better" representation of
performance than Don's example would be one that involved disk access and
network access.
Amigaone v Peg Benchmark : Comment 27 of 177ANN.lu
Posted by DaveP (195.92.67.75) on 08-Mar-2003 09:07 GMT
In reply to Comment 25 (Ole-Egil):
Compiling is a stupid test of performance anyway IMHO.

What we need is peak and load average for different classifications ( which is what spec is going for ) but given that even the big guys have failed to deliver on that so far I won't overburden you with a performance test design.

What you have done so far tells a story. Though those that dislike the story
are apt to dismiss it.
Amigaone v Peg Benchmark : Comment 28 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 09:13 GMT
In reply to Comment 25 (Ole-Egil):
4 out of 5, not 1 out of 5. I actually wrote 5 out of 6, corrected to 4 out of 5, then "corrected" to the wrong line :-)
Amigaone v Peg Benchmark : Comment 29 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 09:22 GMT
In reply to Comment 26 (DaveP):
Hey hey again :-)

I'm not saying you came down on us personally, I just decided you weren't ANY more helpful than ANY of those who dismissed it right away. I appreciate ANY useful input, but "that's not gonna be much better" really wasn't helping :-)

I think I'm fishing for a "this WOULD be good as an additional benchmark".

And to all those who feel dnetc is NOT a benchmark. Well, I disagree. At the moment it is testing two WILDLY different mathematical problems. Sadly it's two problems that BOTH happens to score better on AMD than Intel, and even better on Motorola/IBM then both x86 families... I really wish distributed.net would find some more projects to work on for shorter lengths of time (you know at the current rate we'll take 700 years or something to finish the current project? Optimistic at best :-)).

RC5 is actually widely used at the moment as a means to encrypt information. The rate at which a machine can encrypt/decrypt information is NOT important, you say? But what if that machine IS used to encrypt/decrypt a lot of information, then?

Some tests we could try:

gzip/bzip2 to/from a BIG file that will compress about 75% or so (tests disk io AND cpu throughput)

MD5 on same file. Would test data integrity and disk IO (of course, we would have to distribute this big file in some way, and ensure that it is in fact EXCACTLY the same on all systems, without ANY bit errors. Kinda tough).

hdparm -T -t (this should be uploaded asap, as it's a really simple test we can run in no time, and tests both memory and disk throughput)

So, care to give me more examples?
Amigaone v Peg Benchmark : Comment 30 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 09:29 GMT
In reply to Comment 29 (Ole-Egil):
Hmm, ok, so hdparm is stupid because we're running different disk drives (and I only have ATA66 while all the others have ATA100...)
Amigaone v Peg Benchmark : Comment 31 of 177ANN.lu
Posted by DaveP (195.92.67.75) on 08-Mar-2003 09:32 GMT
In reply to Comment 29 (Ole-Egil):
"I'm not saying you came down on us personally, I just decided you weren't ANY more helpful than ANY of those who dismissed it right away. I appreciate ANY useful input, but "that's not gonna be much better" really wasn't helping :-) "

I can't find where I said "that's not gonna be much better". I referred to
Dons benchmark example alone as needing to add more factors to be representative
of system performance.

I could be wrong, I could have been that cretinous and have just forgot. Please point it out to me because brain is clearly in "skip" mode in that case.
Amigaone v Peg Benchmark : Comment 32 of 177ANN.lu
Posted by pitur (213.224.83.78) on 08-Mar-2003 09:43 GMT
In reply to Comment 8 (Megol):
"For the same money one can buy a AmigaOne G4 XE one can get a dual AMD xp2400+ with 1GByte RAM and a 120GB hdd! "


A dual?? If that's true, could you please post a link to where I can order them, I could afford a renderfarm built out of those...
Amigaone v Peg Benchmark : Comment 33 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 09:47 GMT
In reply to Comment 31 (DaveP):
I didn't quote you, I just say you wasn't helping. And by that I mean when you commented on Don's suggestion. We don't need people to tell us what benchmarks would NOT indicate this or that, we need people to give us _new ideas_. Come on, lighten up already. :-)
Amigaone v Peg Benchmark : Comment 34 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 09:59 GMT
In reply to Comment 32 (pitur):
Hmm, a terawatt renderfarm, eh? :-)

Nah, just kidding. But a farm of those would need some pretty decent cooling.

Render farms these days are blade servers, due to smaller size. Or at least 1U rack units. And neither of those are in the price range mentioned. Both the SE and the XE can be used in 1U render farms, though. I tried to shop around for a 2U MP 2400 system with 1G mem and 120G disk (like someone mentioned here). Motherboard, casing, memory and disk came out at $1700.

And it wouldn't even run AmigaOS 4 :-)
Amigaone v Peg Benchmark : Comment 35 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 10:01 GMT
In reply to Comment 34 (Ole-Egil):
Hmm, god damn me for looking at 1U casings. Now I'm more or ready to drive into town and buy one :-)
Amigaone v Peg Benchmark : Comment 36 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 10:04 GMT
In reply to Comment 25 (Ole-Egil):
Unfortunately all the relevant benchmarks I know of aren't Linux programs so I can't make any specific reccomendations.

I am interested in memory and peripheral bandwidth tests. Ideally one coule have both boards available and simply swap the hard drive and video card between them.

A good test is AGP memory bandwidth (using the same model card) and RAM bandwidth test (copying large blocks of locked memeory).
Amigaone v Peg Benchmark : Comment 37 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 10:06 GMT
In reply to Comment 29 (Ole-Egil):
Compressing data isn't going to burden disk or memory i/o, only CPU. I wonder where you get such ideas |-\
Amigaone v Peg Benchmark : Comment 38 of 177ANN.lu
Posted by Niclas Aronsson (213.67.63.232) on 08-Mar-2003 10:07 GMT
I don't get this. People shout out "WE WANT BENCHMARKS!!!" and when they get it there like "This is soo stupid" JUST because they didn't like the result!

I think this is a nice project! I don't realy care about benchmarks but hey why not if someone have made one =)

And to all complaining they are using the same exe files on all (not the x86) machines. so whats the problem ?!? This is a great way to show the diffrense betwen the boards.

Tnx again
//Niclas
Amigaone v Peg Benchmark : Comment 39 of 177ANN.lu
Posted by DaveP (195.92.67.75) on 08-Mar-2003 10:08 GMT
In reply to Comment 33 (Ole-Egil):
OK so what help do you need? If telling you that including disk i/o and networking i/o into measurements should be added although that does give you
more of an indication of the OS and not the hardware unless the OS is the same
on both hardware ( which it is in this case ).

There are specmark benchmark apps for this, have I got to go and dig out the source code and send it to you?
Amigaone v Peg Benchmark : Comment 40 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 10:10 GMT
In reply to Comment 37 (strobe):
Uhm, it depends on the data and the algoritm, doesn't it?

And I don't see you giving helpful advice, only sour comments. Afraid of being creative, are we?
Amigaone v Peg Benchmark : Comment 41 of 177ANN.lu
Posted by Ole-Egil (195.1.135.34) on 08-Mar-2003 10:10 GMT
In reply to Comment 39 (DaveP):
Yep :-)
Amigaone v Peg Benchmark : Comment 42 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 10:11 GMT
OK, since I'm really interested in seeing this through I researched various benchmarking utilities. This one looks very promising for testing memory:

http://www.bitmover.com/lmbench/
Amigaone v Peg Benchmark : Comment 43 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 10:16 GMT
BTW what I'm really looking for is a Linux equivalent to this app:

http://www.coolmedia.de/throughput/
Amigaone v Peg Benchmark : Comment 44 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 10:28 GMT
Does Linux/PPC have an OpenGL accelerated xserver? Perhaps an OpenGL benchmark could test this.
Amigaone v Peg Benchmark : Comment 45 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 10:30 GMT
In reply to Comment 44 (strobe):
Accelerated GLX in other words.
Amigaone v Peg Benchmark : Comment 46 of 177ANN.lu
Posted by Sam Smith (62.64.210.62) on 08-Mar-2003 10:33 GMT
In reply to Comment 9 (Anonymous):
I think you have to use a little bit of initiative and read between the lines.

The comments over the Teron G4 (no-one mentioned which speed G4 either) being slower than the Pegasos G3 were made clearly as a direct comparrison to the AmigaOne boards. It's just Genesi's little 'dig' at the AOne being a Teron board and not a 'real' Amiga.

The point is - Teron board or not - we just want to run OS4. Some people seem to forget this.

You may wish to believe otherwise but all of this G3 is better than G4 'nonsense' (a word you seem to like ;) was started at almost the same time that the original 'there will be no G4 for Pegasos I' news was released. This is very basic 'marchetecure' and it is pleasing to see that some people do have the ability to see through it. As for asking which G3 and which G4 - there are only a handful of different selections anyway - it was the craziness of the original - G3 is faster than G4 comment that seemed most silly.

---
Sam
Amigaone v Peg Benchmark : Comment 47 of 177ANN.lu
Posted by strobe (12.225.81.107) on 08-Mar-2003 11:02 GMT
OK, forget OpenGL.

All one needs to test AGP bandwidth is x11perf and similar gfx cards.
Amigaone v Peg Benchmark : Comment 48 of 177ANN.lu
Posted by JoŽl EHRET (80.9.104.27) on 08-Mar-2003 11:28 GMT
tests only bangs CPU, not the rest of the environment... do you won't see the difference between patched northbridges and not patched ones.
Moreover, please add the FSB of each machine...because, i know that the pegasos used has 100Mhz FSB, but no idea about the amiga ones... i suppose that the 600 and 666 Mhz A1 used were 133Mhz... because you can't do 666 on a 100Mhz FSB.. :)
And what about the RAM... did you use exactly the same modules... because ther's difference between hi speed kingston ECC rAM and others... so these benchmarks are true, but we need lots of information to analyse them correctly..
Amigaone v Peg Benchmark : Comment 49 of 177ANN.lu
Posted by DaveP (195.92.67.75) on 08-Mar-2003 11:32 GMT
In reply to Comment 48 (JoŽl EHRET):
So why is the Pegasos G3 resoundly beaten by the A1G3 or am I reading the
output wrong?
Amigaone v Peg Benchmark : Comment 50 of 177ANN.lu
Posted by Ben Hermans/Hyperion (134.58.253.129) on 08-Mar-2003 11:36 GMT
In reply to Comment 49 (DaveP):
Yes, strange is it not?

I do distinctly recall somebody claiming that a G3 equipped Pegasos was faster than a G4/800 AmigaOne.

Now I notice that it isn't even faster than an AmigaOne with the same G3.

In fact, it is slower.
Anonymous, there are 177 items in your selection [1 - 50] [51 - 100] [101 - 150] [151 - 177]
Back to Top