A standard CPU

[maxthepadewan2]

Hey Does anyone have any thoughts on what a standard CPU would be around Christmas? I'm a tafe student in Australia, and i was given an assignment recently which requires me to put together two PCs for a group of game developers that will be standard at christmas. I was thinking close to the two gigahertzs mark. Also are their any extra considerations between AMD and Intel, i remember hearing about AMD having some sort of extra programing capabilities with their CPU's is there any truth to this?

PS is there some site i can go to to see comparisons of AMD and Intel as i want to be able to see what the equivelents of each CPU are

 

[Crashman]

Generally speaking, the "leading edge" gamers believe the AMD to be a better performer, something that hasn't been true for months. But still, to get any respect with them, you have to have an AMD CPU system.

Having said that, I think the best CPU value right now is the Intel P4 2.4GHz.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[Col_Kiwi]

Not all gamers are lying to themselves, I've been a hardcore gamer for years and I'm currently solidly supporting P4 over AXP.

-Col.Kiwi

 

[Crashman]

Wow, do you take that thing to LAN parties?

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[imgod2u]

LAN parties? Does anyone actually go near those dorkfests? I think they eat you alive for having an ATI card instead of nVidia.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[Crashman]

Yep

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[maxthepadewan2]

What would be AMDs equivelent to the P4 2.4Gz

 

[imgod2u]

I'd say a 1.9ish GHz Athlon would probably, on average, be equivalent to a 2.4 GHz P4. Give or take.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[halkebul]

"What would be AMDs equivelent to the P4 2.4Gz"

AMD 2400+, it should be available in a few weeks or so.

<i>It's your world kid!!!</i>

 

[lhgpoobaa]

hmmm
really depends which ram platform u put the P4 on. and weather its 400fsb or 533.


<b>Plaid will NEVER go out of style! </b>

 

[lhgpoobaa]

well if you want a fast stable system that costs a minimum of $$ i reccomend something like:
XP1600+ to XP2200+ depending on budget.
via kt333 or sis 735/745 mobo
512mb PC2700 ram.
nvidia Ti4200 graphics card
80Gb hdd.

And if you want some hometruths
a. AMD processors are better at heavy FPU calculation programs.
b. programs optimised for SSE2 go well on the p4.
c. P4 runs better with faster ram. faster the better, but faster = more expensive, especially top of the line PC1066 rdram. hard to find that in Oz.
d. for most things a XP1600+ will be ample. not the fastest on the block, but still quite zippy and o so cheap.
e. P4's seem to run multimedia stuff better, stuff that loves high mem bandwidth, but once again, only with decent ram
f. NEVER under any circumstances get a P4 with SDRAM.

<b>Plaid will NEVER go out of style! </b>

 

[halkebul]

It seems as though this standard will be used for making a critical judgment? For example, the game developers create an executable, then execute it on the standard PC (equipped with standard CPU) and from tests run on that PC (i.e. in-house benchmarks derived from the game's engines, graphics engine for example) they make a decision as to whether they need to further tweak the code and to what degree. But I don't think your professor wants you to supply a standard GHz number, assuming that you attend a university/college with professors that require you to really think about questions before you answer them (MIT for example). Rather, I think he/she wants you to supply standard hardware but with the potential to be dynamic. It would be a good idea to setup one computer with a AMD XP Thoroughbred-B cpu and unclock it (a fellow at HardOCP successfully unlocked a AMD Thoroughbred-B cpu using a mechanical pencil). In that case you could alter the cpu multiplier which would allow you to obtain test results for XP 1500+ (and much lower) through XP 2800+ (and higher), that is to say, a speed range of ~1.5GHz to ~2.8GHz for a typical x86 cpu (pentium3/pentium4 for example). Note that the motherboard will have to be able to accept 333MHz FSB Thoroughbred-B cpu's in order to run at 2700+ and 2800+ levels. It is difficult to determine what the second computer's cpu should be as the first computer takes care of most of the cpu clock speed spectrum. Ah, yes, I would say both computers should contain the AMD Thoroughbred-B cpu but one will have low and high graphics capability (integrated graphics and ATI 9700 Pro in the AGP slot for example) and the other would have midrange graphics capability. The Leadtek GeForce4 Ti 4200 has been shown to perform at GeForce4 Ti 4200/4400/~4600 levels. This should take care of most of the graphics performance spectrum - high, low and mid-range.

Final Recommendation
==================
Computer 1:
AMD 2400+ Thoroughbred-B cpu
nForce2 motherboard with IGP (Integrated Graphics Processor)
ATI 9700 Pro graphics card
(2) PC3200 DDR SDRAM modules

Computer 2:
AMD 2400+ Thoroughbred-B cpu
nForce2 motherboard (with or without IGP)
Leadtek GeForce4 Ti 4200 graphics card
(2) PC3200 DDR SDRAM modules

<i>It's your world kid!!!</i>

 

[why]

I would have to concur. If you are looking to have the standard which are used in the real world and not the standard for sale, then you are looking at a fairly slow system. Probably even slower than 1Ghz on average.

 

[Col_Kiwi]

Yep

Not only will I take this beast <b>to</b> LANs, i host smaller lan parties with friends . I actually only got this system 6 weeks ago and there hasn't been a LAN since, but my last system was also intel based, and i sure as hell used it

Its ironic that those of my friends who have more advanced hardware knowledge and also play games as a hobby respect my system, and some who are more serious gamers but don't know much about hardware think i made a bad choice...

the one thing in common among all those i know (most of which fall into one of the two categories above) is that everyone seems to thing ATI = evil. I try to avoid bias when helping other people, and can respect efforts like R300 for being quite powerful, but in my own systems I tend to avoid ATI becuase of a personal dislike. Back in the days of Rage 128 the drivers and support were so horrible that I still have a personal dislike for ATI.

-Col.Kiwi

 

[chuck232]

The P4 2.4B w/ PC1066 RDRAM beats the XP2400+ which is 2GHz IIRC. It also beats the XP2600+ in many tests. Now you can take this anyway you want. Perhaps with a nForce2 board, the results could be different.

By this Christmas, I'm sure the 2.4B/XP2400+ will be mainstream to only a bit above mainstream. We might have DC DDR too for P4. That would be really awesome. And a R9500 / tuned down NV30 for the VC.

...And all the King's horses and all the King's men couldn't put my computer back together again...

 

[maxthepadewan2]

Thanks for the reply halkebul; you've given me some great ideas. My only after thought is the difference between testing on P4 CPU's and Athlon XP CPU's, I can remember a discussion a few months ago in the class about the different aritectures of AMD and Intel. If I can remember properly, some time ago AMD Developed some sort of extra programming capabilities for their processors, which were a big benefit to programmers. The only drawback to these extra abilities was that they were incompatible with Intel CPU's, and required multiple versions of the game to be created. I was curious as to whether this is still an issue.

This also brings me to another question. If I do what your suggesting, and unlock an Athlon XP 2400, it will supply the developer with an extensive range for his testing, but only for AMD, what about the P4? I hear all the time about how games look really good on promotional computers because the Game has been optimised for that PC. I was wondering if the game would be optimised differently if the processor were a P4? This opens up a lot of extra questions for consideration, I can see you kept your recommendations similar for a reason, Making one of the processors a P4 will make it harder to compare results between the systems, but I need to know if it's necessary to factor the P4 in somewhere.

Summing up - would it be necessary to use a P4 for testing, considering the different architectures of AMD and Intel CPU's? Can you unlock a P4 like the Athlon you were talking about? And would using a P4 considerably influence the comparing of results between the systems?

Thanks to everyone for their time.

 

[halkebul]

It would not be possible for a single P4 to mimic a wide range of P4s because they are unlockable which means you can't change the multiplier. You can only change the clock speed of the P4 by changing the FSB but this results in unofficial P4s. For example, a 1.8A(100MHz FSB, 400MHz quad-pumped) cannot mimic a 2.2A (100MHz FSB, 400MHz quad-pumped) because although the 1.8A can reach 2.2GHz it would have to do so by increasing the FSB.

2.2A (100MHz FSB at default) at 2.2GHz = 22 (multiplier) x 100MHz (FSB) = 2.2GHz
1.8A (100MHz FSB at default) at 2.2GHz = 18 (multiplier) x 122MHz (FSB) = 2.2GHz

The FSB speed difference is too high for us to consider the two reasonably equivalent. The results are also unacceptable when ya try to mimic higher speed P4s (2.5GHz) and lower speed P4s (1.5GHz). Plus an official P4 1.5GHz has less cache (256K) than a 1.8A (512K) - a different core altogether.

The Thoroughbred-B Athlon, on the other hand, gives us better approximations of different processors because we can change the multiplier (XP unlocking is greatly simplified by getting a Athlon XP unlocking kit, $12 at highSpeedPC.com). That is, a 2400+ can be changed into a 2600+ by changing the multiplier of the 2400+ from 15 to 16. A 2400+ can be changed into a 2800+ by changing the multiplier of the 2400+ from 15 to 13.5 and the FSB from 133 to 166. And a 2400+ would be nearly equivalent (i say <i>nearly equivalent</i> as opposed to <i>equivalent</i> because the 2400+ Thoroughbred-B and 1500+ palomino have different cores, same amount of cache though between the two cores) to a 1500+ if ya change the multiplier of the 2400+ to 10.

You can make comparisons to Pentium4s also. A AMD 1500+ is nearly equivalent to a P4 @ 1.5GHz, AMD 2800+ is nearly equivalent to a P4 @ 2.8 GHz. And so on ...

A P4 at 2.4GHz may considerably, with respect to equivalency testing that is, outperform a AMD 2400+ in SSE2 opimized applications because P4 has SSE2 support and Athlon does not. But don't worry about this because SSE2 is not usually found in games. Happily, none of the Athlon or Pentium features require one to produce two versions of software.


<i>It's your world kid!!!</i>

 

[maxthepadewan2]

Thanks again for the help halkebul.

I had a talk to my teacher about providing CPU's that are unlockable (he was happy I had given it the extra thought, and I was forgiven for providing the project late, so thanks for the help). He does like the idea as it gives him considerably better testing options. When I asked him if I should also consider the P4, he stressed that I should probably cover all my bases. I'm considering adding another pc optimised around the P4 for testing purposes.

I should probably explain a bit more regarding what the project is all about. So I can explain my motives a little better. I'll give you the outline.

The teacher that issued the assignment designed a game a few years ago (the game ended up being sold as a psychological tester for a major airline company). I think he was trying to give us something fun to do, so we'd get a little more motivated to learn. Anyway, Les has given us a scenario where all the old games team have got back together to make another game. We have to provide all the equipment to make this possible just like if it was a real business proposal.

There are only five members of the team; - a background artist, a character artist, a soundman, a C++ expert and an AI specialist. The first three members of the team all need state-of-the-art equipment, the last two need pc's that will most likely be purchased at Christmas (so the game can be optimised for the most common computer(s) when it’s released). We initiated the project with an E-mail and were given basic guidelines as to what the team needs but basically it’s up to us to use our best judgement.

What I was considering, was adding an extra pc specifically for the teams testing purposes, and supplying them with some of the most common components - like extra motherboards and chipsets, different ram, hard drives from maxtor and seagate and extra sound cards like the SB live etc. If necessary, I could probably add some of the different P4's CPU's as well so I can cover all the different performance areas and even the different heat and performance issues of different cores. What do you think? Am I going a bit overboard? Money is not really an issue provided I can justify.

Going to another part of the project, I was wondering if you had any thoughts on using Dual Intel P4 Xeon 2.2 GHz CPU's for a high end graphics station. They'll be coupled with the best high end wild cat card from 3Dlabs, (which costs over $3000 US.

I'd go over a few more of the details for the gamer’s project but I'm not really sure if it'd be appropriate in a CPU forum.

I've had an after thought. I re-read your last post, when I came to the part about the different sizes of cache with different processers it got me thinking about how this will effect performance. I remember a discussion in class about how the different sizes of processor level one cache need to be considered in the programming of a game and how the programmers had to consider this . Do you think I should include a Duron and a Celeron processor for testing as well? as these are pretty common, and provide a better idea as to how the game will perform with different sizes of processor cache.

 

[halkebul]

Going to another part of the project, I was wondering if you had any thoughts on using Duel Intel P4 Xeon 2.2 GHz CPU's for a high end graphics station. They'll be coupled with the best high end wild cat card from 3Dlabs, (which costs over $3000 US.

My recommendation for 3D-graphics (I assume its 3D games) development at Dec 2002 would be:
ATI FireGL X1 Graphics Card
Dual 2.8GHz XEON processors

The Wildcat cost more but nowhere near as good as the <A HREF="http://mirror.ati.com/products/workstation/fireglx1/index.html" target="_new">ATI FireGL X1 graphics card</A> (available soon). Its the fastest and the best. Note that you would need an Operating System and Software that supports dual processors to take advantage of the second processor otherwise the second processor would not be utilized. Windows XP Professional Edition supports dual processors while Windows XP Home Edition does not. 3D Graphics Programs such as 3D Studio max, Maya, Lightwave and the like all take advantage of a second cpu.

<i>It's your world kid!!!</i>

 

[halkebul]

I had a talk to my teacher about providing CPU's that are unlockable (he was happy I had given it the extra thought, and I was forgiven for providing the project late, so thanks for the help). He does like the idea as it gives him considerably better testing options. When I asked him if I should also consider the P4, he stressed that I should probably cover all my bases. I'm considering adding another pc optimised around the P4 for testing purposes.

and

What I was considering, was adding an extra pc specifically for the teams testing purposes, and supplying them with some of the most common components - like extra motherboard boards and chipsets, different ram, hard drives from maxtor and seagate and extra sound cards like the SB live etc. If necessary, I could probably add some of the different P4's CPU's as well so I can cover all the different performance areas and even the different heat and performance issues of different cores. What do you think? Am I going a bit overboard? Money is not really an issue provided I can justify.

If ya need to cover all areas, the P4 2.4B (2.4GHz, 533MHz FSB) would be the best option. I give this recommendation only because your professor suggested it as the other two testing systems, equipped with unlocked AMD Thoroughbred-Bs, already fulfill most of the cpu performance spectrum. However if ya have the extra money, ya might as well have some AMD Durons, Intel Pentium2s and Pentium3s (and supporting motherboards and RAM) laying around just because ya can and they might be useful. And some other graphics cards, no matter how ancient. Hey, its christmas!!!

<i>It's your world kid!!!</i>

 

[halkebul]

The teacher that issued the assignment designed a game a few years ago (the game ended up being sold as a psychological tester for a major airline company). I think he was trying to give us something fun to do, so we'd get a little more motivated to learn. Anyway, Les has given us a scenario where all the old games team have got back together to make another game. We have to provide all the equipment to make this possible just like if it was a real business proposal.

From this, I am thinking that this new game will be based on the old game but with modern day graphics (because of faster graphics card), AI (because of faster CPUs) etc. Would I be correct in assuming this?

<i>It's your world kid!!!</i>

 

[halkebul]

I've had an after thought. I re-read your last post, when I came to the part about the different sizes of cache with different processers it got me thinking about how this will effect performance.

The performance difference is negligible for testing purposes.

<i>It's your world kid!!!</i>

 

[halkebul]

There are only five members of the team; - a background artist, a character artist, a soundman, a C++ expert and an AI specialist. The first three members of the team all need state-of-the-art equipment, the last two need pc's that will most likely be purchased at Christmas (so the game can be optimised for the most common computer(s) when it’s released). We initiated the project with an E-mail and were given basic guidelines as to what the team needs but basically it’s up to us to use our best judgement.

Revised Recommendation:
Two unlocked Thoroughbred B Computers for testing
P4 2.4B computer for testing
2 graphics computers (ATI FireGL X1 and Dual 2.8GHz xeons) for development
3 computers for sound/AI/c++ (3.06GHz P4) for development


<i>It's your world kid!!!</i>

 

[maxthepadewan2]

I've assumed the game will be based on a 3D environment myself, (even though the last version of the game was most likely based around 2D images) as the developers are aiming to produce a professional quality product, which will compete with current games.

I've decided to stay with the 3d Labs Wildcat III 6210, the ATI FireGL X1 Graphics Card looks impressive, but I was unable to find an estimate of a price and the card needs to be available, (plus I’ve already researched the wildcat and the parhelia and I’m running out of time)

 

[maxthepadewan2]

"From this, I am thinking that this new game will be based on the old game but with modern day graphics (because of faster graphics card), AI (because of faster CPUs) etc. Would I be correct in assuming this?"

You would be correct in assuming this, though I'm unsure of the particulars for how the new game will be designed. The Information I received from Les does acknowledge that some of the older material may be used in the new game. Unfortunately you've brought worrying information to my attention, as this does make it more likely that the game will be based on 2D Images after all.