Fx5800U?!?! Plz help

[ChrisRay]

Originally Posted by quik_2_win

Indeed- the 5800U is silent during 2D operation/ quite noisy during 3D. As far as benchmarks, the 5800U excels because of raw fill-rate- a byproduct of higher clockspeeds...5900 brought nothing new to the table (save for a 256-bit memory bus that had negligable benefeits). Even running shadermark shows the advantage of the 5800U over it's successor-
Here is a 5900 (courtesy of Anandtech.com)

neither card is modified or overclocked AFAIK
...and here is a (my) 5800U:

Either Anti Detect isnt working properly. Or there is something "seriously" wrong" with anandtech results. I dont know. But Here's my results. And they look quite a bit more impressive.

Forgive the compression, I needed to fit the file down so I could upload it.

These results are from my 5900 @ 465/900



Anyway, Whether people want to believe it or not. The FX 5900 Does have twice the Floating Point capabilities of the a 5800, the 5800 has two integer units, And the 5900 series had those 2 units replaced by 2 FP32 units, Which can divide themselves into Quad FP16 operations.


*edit* Just realised that was anandtechs results. Not beyond3d.

[muzz]

Chris how much of a difference does the 256 vs 128 make on the NV cards? 5900 is 256 right vs 5800 128
And what were his system specs in relation to yours?

[ChrisRay]

Originally Posted by muzz

Chris how much of a difference does the 256 vs 128 make on the NV cards? 5900 is 256 right vs 5800 128
And what were his system specs in relation to yours?

In Shader intensive apps, The 256 Memory Bus doesnt do much for the 5900, It becomes shader limited long before it becomes bandwith Limited. Hope that helps you.

He said he was running a 5800 Ultra, So I can only assume his results are @ 500/1000

[quik_2_win]

Originally Posted by ChrisRay

...something "seriously" wrong" with anandtech results...

There may well be, I have no Idea

Here's my results... my 5900 @ 465/900

The numbers look good, however, keep in mind that neither of the other cards were overclocked. Assuming the bench would scale with clockspeed, the other cards may well have scored 5-15% higher...

... The FX 5900 Does have twice the Floating Point capabilities of the a 5800...

I was not aware of that, but from looking at your numbers, I would certainly say the 5900 has some enhancements that show-up in shadermark. While not night-and-day, the improvement is there.

[ChrisRay]

Originally Posted by quik_2_win

There may well be, I have no Idea

The numbers look good, however, keep in mind that neither of the other cards were overclocked. Assuming the bench would scale with clockspeed, the other cards may well have scored 5-15% higher...

I was not aware of that, but from looking at your numbers, I would certainly say the 5900 has some enhancements that show-up in shadermark. While not night-and-day, the improvement is there.

Well the Only reason I didnt drop my clocks to stock speed. Was because an FX 5950 is clocked @ 475/950, so I figured I'd be my results would be less, Than that of a 5950.

I didnt mean to sound patronizing. when I read my paragraph, I may have sounded that way, I am terribly sorry if I did.. I assume a 5950 can scale to about 500 mhz and Not sure about a 5800 Ultra, Perhaps you can tell me But my results should be at least somewhat comparative to a 5950 which is clocked 10 mhz faster and 50 mhz faster on Memory.

But yes, the 5900 series had the two integer units replaced with a set of FP32 units, Which considerably helped FP performance.

Anywho,

[Lfctony]

Well, in my own experience, I've been talking with a customer that purchased his system from me, half a year ago. The system is a P4 2.6C with 512MB DC DDR400, and a 5800. Its at stock speeds and not one single game has given him trouble so far. Farcry was working horribly and he called in for support. It turned out he was using one of the earlier 45.23 drivers. I told him to swap to the 56.72s and after doing so, he is currently playing the game at max settings @ 1024x768.

My point is, the 5800 is a good card, the 5900 is even better, if an aging 5800 works well in all games today, you'll definitely have no problem with a 5900XT.

[Quasar]

My personal opinion on this matter, as i am using a FX5800U on a 1,4GHz P3:

- undetecable (ear-wise) in 2D (but beware of VMR-Mixing, this uses 3D-Mode)
- barely bearable in 3D, unless you do not play your games quite loud.
- very high fillrates for older games
- not much use going for cheat-wares Driver above 45.23 'cause it does not profit that greatly from FP16-Shaders as does nV35.
- also, with 45.23 or older Dets you get full filtering (trilinear/AF on all stages 'cept UT2003).

Nice thing, but Far Cry drives it to it's limits, in pure PS1.1-Mode though, almost as fast as an R9800XT even in Far Cry

[Kamel]

maybe he meant 6800

[Xplode]

JUST GOT THE VID CARD...

Installed it just fine... I'm very impressed with it, jump from 800x600xmed raven shield to 1280x1024xhighestx4aa/8af. yes quite impressive,Would overclocking help me gain alot of FP in games? cause coolbits Auto Detect was 465/860 even tho I expected more since people said this thing has 2.2ns memory and will go up over 900... Gonna try 3dmark03 now, my prevouis score was 1525 with my Fx5200. Also I had to overwrite 56.72 drivers, would it be better if I uninstall them,then install em again?

Xplode

[ragejg]

yeh, uninstall, then install again... but by the sounds of your ravenshield results, everything looks okay...

That's too bad about your memory OC tho... have you tried manually setting it @ 900?

[Xplode]

No havnt tried yet, but I'm pretty sure it'll let me go there. With my Fx5200 I had the speed at over the Auto Detect point. I just dont wanna fry my card and go to far thom still got 2 months till my new rig... Any recommendation on overclocking this badboy?

[ragejg]

475/900