PureBasic Forums - English

i dont know of any of you have heard of performance test, software that runs a series of tests on your machine like floating math, drawing, memory tests, then gives you a rating. the rating can be compared with other machines in data files.

now...

a pentium 4 3.1 Ghz 512 cache hyperthreading with 512 megs DDR RAM outperforms my dual Xeon 3.06 Ghz 512 cache hyperthreading with 2 Gigs RAM and 256 megs video, especially in the processor tests. how can that possibly happen??????? it takes 1 min for all tests to run, which is very very fast.

my machine should be able to process two to 3 times faster than a single processor. are the tests inaccurate?

on another note, i have yet to find a single game, graphics program, or software that can remotely slow my machine down. are there any graphics programs out there that can really test its abilities? ive even tried solidedge 10, working with a part assembly that requires 800 Megs memory, and it didnt even flinch. any ideas?

Maybe the tests are only utilizing one processor? Have you tried running two coppies of the test at once?

I'm not very knowlegable in this area obviously

> i have yet to find a single game, graphics program, or software that can
> remotely slow my machine down. are there any graphics programs out
> there that can really test its abilities?

From what I've read, Doom3 (even the demo) should give your machine
a real good pounding... make sure you turn ALL its graphics to "High".

Or, try running an emulator (such as WinUAE [the Amiga]) and run a "big"
Amiga game that used to run slow on real Amigas, such as "Alien Breed 3D",
or its sequel, "Alien Breed 3D: The Killing Grounds". It'd be interesting to
see if such games run at real-time on your PC under emulation.

BTW, I'd love my PC to be able to handle anything... you sound like you're
disappointed by it...?

localmotion34 wrote:a pentium 4 3.1 Ghz 512 cache hyperthreading with 512 megs DDR RAM outperforms my dual Xeon 3.06 Ghz 512 cache hyperthreading with 2 Gigs RAM and 256 megs video, especially in the processor tests. how can that possibly happen??????? it takes 1 min for all tests to run, which is very very fast.

my machine should be able to process two to 3 times faster than a single processor. are the tests inaccurate?

1. A Dual-Processor machine running a program that utilizes MPS is never 2 times as fast as a similar Uni-CPU PC due to some overhead in managing 2 processors

2. A program that doesn't utilize both CPU is slower on that machine as on a similar PC.

3. Usually, Xeon processors perform better with HT disabled, while P4 perform better.

4. Memory performance of Xeons is lower compared to an equal P4.

So, as long as you don't use the appropriate software which supports MPS, you are better of with a a single P4, compared to a single- or dual-Xeon. In case you use the appropriate software (or have lots of stuff running at the same time), then the dual-Xeon machine will perform better - but never at twice the speed of a single-Xeon and even a bit less compared to a P4.

depending on your point of view, ht just uses the spare capacity of your processor... (parts of the cpu not in use for a specific command at a specific time) so it only helps multi threaded applications, or multiple applications

a single thread will in concept not run any faster on a ht cpu (ok, a little due to the os stealing timeslices, but then again that os IS another thread)

ht makes a system that is busy with something feel more responsive, true

a system with multiple cpu's at low load will perform similar to a single cpu with ht

when the (background) load increases the multi cpu system will, however, pick up and perform better

interestingly enough, the non-ht amd athlons (xp) are appearently a lot more efficient in code execution, in spite of their lower clockspeeds, try multiple threads on (performance wise) similar athlons and p4's, and the athlon doesn't do a bad job at all

however, at these high speeds performance is hard to express in absolute numbers, unless tailored to a specific application (transactions on a database, pages per minute on a webserver etc.) i.e. when you're dealing with a client or stand alone machine the subjective perception of the user is more important, in other words: how does it *feel*

no matter how fast the cpu, if all windows xp's gadgets and services are switched on, a virus scanner, registry monitor, temperature monitor, personal firewall, memory manager, monitor manager, virtual cd emulator, and some messenger are running simultaneously, the whole hardware and software configuration together defines how fast the machine feels, add a little spyware and your glorious dual whatever machine starts crawling (ok, it crawls fast, i admit that )

i'm now using a tweaked windows xp with most unnecessary services of but all those tools mentioned above on, on a homebuilt xp2500 / 1 gb / asrock mb / geforce 4 / 2x 120 gb hdd, and it runs quite nice, however i've seen a few (standard) (a-brand) boxes that were outright pathethic, even factory configured (packard bell, hp, sony)

forget the absolute numbers, how does the machine *feel* is what counts for a user (not for a server)

it seems like i should not buy a big 2 processor machine then.
My current system is pretty good, but for free i just got an old dual 266 mhz, and actually that perfoms pretty good in windows 2000. better than a 500 mhz.
with 256 mb ram hehe, but its fast enough to at least do something.

what hardware / cpu's?

i've been playing around a little with 'hardware on the edge', see here...

http://www.xs4all.nl/~bluez/datatalk/lowgo.htm

if it's a dual pentium pro with a decent harddrive (transfer and track access) it'll run like a rocket if you add a little more memory and kill all unnecessary services

win2k may be the better choice, opinions differ, i slighly prefer xp but not by a large margin

i think its a pentium pro (i know its a pentium, but not if its a pro or p2 or what it is.). i first get it after sunday (havent had time to pick it up yet..), The harddrive is a small 15 gb.
but 256 mb of ram..

I prefer XP too, but isnt w2k a bit faster on low end systems? (but of course if you turn of the visual skin/theme and all the effects in xp, it might run better). about the graphic card, im not 100% sure. Either its an
old Vodoo 2 or a geeforce 2. Maybe its back in the TNT cards, but i belive its one of the other ones.

I think ill go reading your guide when i get it, it seems pretty nice.

a dual pro runs well, check the type of ram before you go to the shop and buy some, some of the a-brands (compaq comes to mind) use special memory modules so you may not be able to use off the shelf stuff

i know of a dual pentium pro with 128 mb running win2k as a server, can't remember how much disk space (iirc some scsi 80 gb drives in a raid config) but it runs fine

re. graphics card: if it's a voodoo or the likes it's probably not an a-brand machine i'd prefer a geforce 2 though...

don't expect the hottest thing in games, but it sure sounds like an okay machine, perhaps you may have to add a better / faster hdd or a little mem depending on your usage (my own experience has shown shown that at 256+ xp suddenly runs instead of crawls, and that disk speed made a lot more difference than i expected)

hi!

actually the voodo 2 cards was amazing. Geeforce 2 is a newer card, therefore better. But i had an voodo 2 when it just came out, and that time it was truly amazing. I had it long time, but i could actually play Unreal Tournament without any problems.

The small computer is not for games. I actually though of it to have some things running so that my main computer wouldnt have to use resources on it. Messenger , so i can chat while playing
also ftp server or whatever i have running.