My New 1.333 GHz AMD System Is A Bit Slow.

[darkshadow1]

Wait...but shouldn't the video card matter? I mean leaving saturation aside...the geforce3 is faster than a geforce2, even at 640X480, no?

Maybe I'm not thinking straight, but if you went by saturation logic alone you could test the setup with a TNT and get the same numbers?

 

[jeffrey]

The scores that you are comparing to won't be the same because:

Anand's test used NVIDIA's Detonator3 v12.41 drivers
Anand's test used a NVIDIA Geforce 3 64mb
Anand's test used 256MB PC2100 Corsair ddr RAM
Anand's test disabled the Sound

Because of this the number of 179 fps in quake III isn't apples to apples.
His setup

your scores look pretty good

 

[gnoymyguy]

the person above is correct

 

[BFG10K]

Yeah, maybe you guys are right. Maybe my scores are OK. My games do run fast and smooth and I don't have any problems whatsoever with my rig. Considering it's my first AMD system, that's pretty good.

It's just that I was expecting something a bit better, especially in UT which is heavily influenced by the CPU.

Wait...but shouldn't the video card matter? I mean leaving saturation aside...the geforce3 is faster than a geforce2, even at 640X480, no?

With today's top of the line CPUs it isn't unless you're using DX8 optimised programs which use programmable T&L.

Maybe I'm not thinking straight, but if you went by saturation logic alone you could test the setup with a TNT and get the same numbers?

No, look: the easiest way to tell if you're video card limited or not is to lower the resolution. If your scores stay the same when you do this, you're CPU limited. If they increase, you're video card limited.

Regardless of whether I run my tests at 640 x 480 x 32 or 320 x 240 x 16, the scores stay the same. That means I'm CPU limited and that's precisely why I chose such settings to begin with. Usually I don't use such low-res settings.

Honestly, they're very basic benchmarking principles.

 

[HappyGamer2]

well i don't know about your system, but on mine with a voodoo5 when I benched UT at 1024x768 max quality in glide i got almost perfect total scaling when I overclocked my cpu, 800, 988 then 1062, frame per cpu were almost equal, actually slight better with fbs increase.
true you not using glide or a v5 but even so I think you should have more incease than you got, you using d3d or opengl? try both and see how they scale

 

[LXi]

After 30 posts we finally realized Anandtech ran their benches with a faster video card, different drivers, different mobo and RAM, and with sound turned off. Hell yea, I think those are enough to make up the 33fps difference. Anyway, congrats on your AMD/VIA system, Im glad everything went smooth aftering hearing you comment negatively on AMD/VIA numorous times.

 

[gnoymyguy]

Lxi - Heehee, so the question is not why the 1.33 ghz AMD System a bit slow, but rather why are WE a bit slow? This could be interpreted as an understatement


and you'd be right

 

[BFG10K]

It still doesn't explain the UT scores, bright sparks. Nor does it explain the people who agree with me about CPU scaling.

Anyway, thankyou very much to everyone who replied.

 

[Weyoun]

I have generally found, in my experience, that the 4.32 4 in 1s possess quite a few demons, especialyl with regard to gameplay and IDE drivers... I had my system running the 4.32 4in1s and the opening scene of Operation Flashpoint was noteably slower than 4.31.... Just make sure that you're in safe mode when you install the damn things... That's the only way i've actually gotten them to change

might be worth a shot, hope this helps

 

[Rectalfier]

Disabling sound is known to boost FPS by 20-30FPS. People used to hammer Sharky Extreme for having lower benchmark scores than Anand.

 

[HappyGamer2]

Sound really has nothing to do with his issue, the issue is the lack of frame rate increase with a more powerful cpu.
it' sound on vs sound on as I take it. I don't think he got the frame rate increase in UT that he should have gotten. quake3 maybe a different story thou.

 

[Dark4ng3l]

Hmmmm I'd say those fps are not low enough to be caused by anything major, probably just little things here and there. You mbob might be slower thna what anand used, the gf3 might have some sort of weird optimising that makes it faster at lower resolutions. wait a sec do you have dram interleaving on?? I know some mobo's dont even have the option to turn it on...

 

[Def]

Your memory bandwidth didn't increase that much. That's why your scores didn't directly scale with the increase in CPU speed.

If your P3 was already using every bit of memory bandwidth it had, then only increasing your memory speed a few Mhz and increasing your CPU speed by hundreds of Mhz isn't going to make much of a difference.

Think of a GF2 GTS, at stock speeds it's held back by the memory bandwidth. If you increase the Core speed, it barely increases framerate at all. But if you overclock the memory, the scores increase dramatically.

BTW, a DDR system would be faster, especially in the low resolution you are benchmarking.

 

[BFG10K]

OK, I'm scoring ~190 FPS in Quake 3 with the sound off so everything's OK. Honestly, I've never seen sound make such a difference in benchmarking scores.

OTOH I'm convinced UT could run faster (sound off gives me 2 extra FPS) but I think it's the SDRAM memory holding me back, like the other guys say. That explains why my scores increased when I turned on 4-way interleaving. I guess I'll just try to tweak the memory in the BIOS a bit more and see if I can squeeze a bit more performance out of it.

I didn't buy DDR RAM because I already have 320 MB SDRAM and I didn't want to get rid of it.

 

[gnoymyguy]

are you using default 1.33x10? if so have you thought about increasing fsb and maybe lowering the multiplier if you want to keep around 1.33?

Personally I found that running 9/9.5x100=900/950 on my duron is not always as fast as 6.5x133=864, the extra bandwidth of the memory making up for the loss of a few mhz