Ace's 2003 Gamers' Hardware Upgrade Guide

[apoppin]

The basic concept of our article is that, starting from a slightly "old" configuration, we investigate which upgrades give you the best return on your investment.

This is a VERY interesting read.

We perform the following upgrades:

Video Card Upgrade:
We upgrade the Geforce 2 Ti 200 to more trendy video cards, such as the Geforce 4 MX 440, Geforce 4 Ti 4200/4600 and Radeon 9700 Pro (Tyan Tachyon G9700). We also check with a Geforce 3 Ti 200 and Radeon 8500, as both videoards featured an excellent price/performance ratio and have been very popular.
CPU Upgrade:
We swap our old Thunderbird for a brand new Thoroughbred Athlon XP.
Motherboard Upgrade:
We simply change the motherboard for one with the fastest Athlon chipset: the nForce 2
CPU + Motherboard Upgrade
Everything:
CPU, motherboard and video card.
Should you get that flashy new 3D card or do you need more CPU power? Let's find out...

 

[RSMemphis]

There's something wrong with their benchmarks. The Radeon 8500 is consistently outdone by a GeForce 4MX or a GeForce 3 Ti200?
What's with that?

 

[dullard]

At times the 2800+ was slower than the 1.4 GHz Athlon with the same motherboard and videocard. That seemed a bit odd.

 

[Darien]

Originally posted by: RSMemphis
There's something wrong with their benchmarks. The Radeon 8500 is consistently outdone by a GeForce 4MX or a GeForce 3 Ti200?
What's with that?

Not consistently. It wins (just by a bit) in the first few benchmarks, but loses (just by a bit) in the last few.

Especially the last one:
http://www.aceshardware.com/read.jsp?id=50000363
Grand Prix 4 (1600x1200x32)
Athlon 2800+ Radeon 8500 28
Athlon 2800+ Geforce 4 MX440 37
Athlon 2800+ Geforce 3 Ti 200 37


Probably showing off the limitation of the Radeon architecture...or driver issue?

EDIT: Linkage

 

[RSMemphis]

Well, maybe they did not clean out the drivers perfectly, i.e. install ATI drivers after having it run NVidia gfx cards.

Dullard, I noticed that too. Very weird.

 

[apoppin]

We can come to two important conclusions. Firstly, it is clear that what we told you a year ago, is no longer entirely valid. In the past, RPG, RTS and simulators were mostly CPU limited, and a fast video card could not boost these games to much higher framerates. As games are using more and more of the new possibilities of pixel and vertex shaders, and are leveraging the power of the T&L unit, the importance of the video card grows. Specifically, many games now feature several levels of detail, enabling those with faster video cards to enable extra shadowing, higher levels of geometry, higher resolution textures, and so forth.

In the more strategy-oriented first person shooters -- namely Ghost Recon and Battlefield 1942 -- we see the opposite trend. In these games, the NPC and enemy AI can make or break the game. The amount of time spent in AI routines has percentage wise grown enormously. Back in 1999, AI demanded about 1-4% of the CPU's power. Today, CPUs are spending up to 25% of their time crunching through AI algorithms. CPU power is thus getting more important for many such games.

The best approach to upgrading is to consider what types of games you play most, and then upgrade the component that will provide the most benefit to that genre. In the case of Battlefield 1942 and Ghost Recon, a new CPU may be in order, while other games may warrant the purchase of a new video card. . . .

😉

 

[apoppin]

Originally posted by: Darien

Originally posted by: RSMemphis
There's something wrong with their benchmarks. The Radeon 8500 is consistently outdone by a GeForce 4MX or a GeForce 3 Ti200?
What's with that?

Not consistently. It wins (just by a bit) in the first few benchmarks, but loses (just by a bit) in the last few.

Especially the last one:
http://www.aceshardware.com/read.jsp?id=50000363
Grand Prix 4 (1600x1200x32)
Athlon 2800+ Radeon 8500 28
Athlon 2800+ Geforce 4 MX440 37
Athlon 2800+ Geforce 3 Ti 200 37


Probably showing off the limitation of the Radeon architecture...or driver issue?

EDIT: Linkage

Oops, I meant to answer this in my last post. 😱
Ace (actually) gives the answer:

Only at 1600x1200, the video card starts to play. And even with a humble Geforce 4 MX440, the game maxes out at 37 fps. Clearly a CPU limited game.

 

[Darien]

Originally posted by: apoppin

Originally posted by: Darien

Originally posted by: RSMemphis
There's something wrong with their benchmarks. The Radeon 8500 is consistently outdone by a GeForce 4MX or a GeForce 3 Ti200?
What's with that?

Not consistently. It wins (just by a bit) in the first few benchmarks, but loses (just by a bit) in the last few.

Especially the last one:
http://www.aceshardware.com/read.jsp?id=50000363
Grand Prix 4 (1600x1200x32)
Athlon 2800+ Radeon 8500 28
Athlon 2800+ Geforce 4 MX440 37
Athlon 2800+ Geforce 3 Ti 200 37


Probably showing off the limitation of the Radeon architecture...or driver issue?

EDIT: Linkage

Oops, I meant to answer this in my last post. 😱
Ace (actually) gives the answer:

Only at 1600x1200, the video card starts to play. And even with a humble Geforce 4 MX440, the game maxes out at 37 fps. Clearly a CPU limited game.

I agree that it's CPU limited, but it doesn't explain why the Radeon has ~9 FPS less than the MX. So drivers?

 

[apoppin]

Originally posted by: Darien

Originally posted by: apoppin

Originally posted by: Darien

Originally posted by: RSMemphis
There's something wrong with their benchmarks. The Radeon 8500 is consistently outdone by a GeForce 4MX or a GeForce 3 Ti200?
What's with that?

Not consistently. It wins (just by a bit) in the first few benchmarks, but loses (just by a bit) in the last few.

Especially the last one:
http://www.aceshardware.com/read.jsp?id=50000363
Grand Prix 4 (1600x1200x32)
Athlon 2800+ Radeon 8500 28
Athlon 2800+ Geforce 4 MX440 37
Athlon 2800+ Geforce 3 Ti 200 37


Probably showing off the limitation of the Radeon architecture...or driver issue?

EDIT: Linkage

Oops, I meant to answer this in my last post. 😱
Ace (actually) gives the answer:

Only at 1600x1200, the video card starts to play. And even with a humble Geforce 4 MX440, the game maxes out at 37 fps. Clearly a CPU limited game.

I agree that it's CPU limited, but it doesn't explain why the Radeon has ~9 FPS less than the MX. So drivers?

Possibly . . . but I don't think so . . . possibly it's "older architecture" holds it back . . . however, Ace's Forum is discussing this article currently and they have also found several errors.

 

[dem0nic]

These guys are idiots.. Whats the point of having a R9700, if you aren't going to a least turn on AA or AF.

Zzzz

 

[Snoop]

Originally posted by: dem0nic
These guys are idiots.. Whats the point of having a R9700, if you aren't going to a least turn on AA or AF.

Zzzz

No, your an idiot. The review wasnt about pure benchmarked speed, it was about upgrading a computer and what parts will give you the most benefit. So, STFU k thx. troll


Also, as for the Radeon 8500 benchmarks showing the g4mx outperforming it, I have read that the radeons drivers are much more highly optimized for the P4 and the Radeons seem to lose some of the advantage they gain when benchmarked with athlons instead of P4's. I would be interesting if someone could take a radeon 9700 and a geforce ti 4600, benchmark them on seperate AMD and Intel systems then measure the performance delta btw them. (ie. the 9700 beats the 4600 by an average of 20% on the P4 system while on the Athlon it only wins by 15%)

 

[apoppin]

Does the "weekend" crowd have any opinions? . . . it's pretty interesting. 😉

 

[RaynorWolfcastle]

the Radeon 8500/GF4 MX issue may have to do with the fact that they used the Cat 3.0 drivers (which I assume are optimized for DirectX 9) while using DirectX 8.1.

Just a thought.

 

[WyteWatt]

apoppin to bad they didn't do sdram vs ddr ram. I wish someone would do sdram vs ddr ram but at a resolution of at least 800x600, 1024x768, and higher but it still was good. Amazing how important your processor is for some games. Some games you won't beable to get 100 fps in intill 5 to 10 ghz cpus come out 🙁 The reason i want 100 fps is so the minimum fps will be about 60 fps. You need some safety net room you know.

 

[Bovinicus]

the Radeon 8500/GF4 MX issue may have to do with the fact that they used the Cat 3.0 drivers (which I assume are optimized for DirectX 9) while using DirectX 8.1.

I don't think there is an issue. Some video cards perform especially well in certain games, and worse in others. That is just the way it is. There are a few benchmarks where the Ti4200 can barely beat the 8500. Neverwinter Nights and Grand Prix just don't seem to like the Radeon 8500. Of course, they don't take into account it's strong point either: Anisotropic filtering. In addition, Neverwinter Nights runs perfectly fine on my system with an 8500. =)

 

[apoppin]

Originally posted by: imtim83
apoppin to bad they didn't do sdram vs ddr ram. I wish someone would do sdram vs ddr ram but at a resolution of at least 800x600, 1024x768, and higher but it still was good. Amazing how important your processor is for some games. Some games you won't beable to get 100 fps in intill 5 to 10 ghz cpus come out 🙁 The reason i want 100 fps is so the minimum fps will be about 60 fps. You need some safety net room you know.

Just expect about a 10% performance 'hit' with SDRAM compared with DDR . . . "about", depending on the game.