ATI HD4890 Owners Thread - Page 11 - Overclock.net

Ricdeau · 04-09-09

Quote:

Originally Posted by BradleyW

how did you get 4890 cf to work. nobody can get cf for those to work. only 1 card is detected for many people

Most of the people I saw having issues were using 9.3's. The 9.4 betas from what I've seen have worked for everyone I've seen using them.

gerikoh · 04-09-09

so anyone can confirm if you can mix crossfire this with a 4870?

MasterKromm · 04-09-09

I've been intently following this thread, it's been a while since I've upgraded my GPU... While my current GPU will do 19x12, 16xAF and 0-4xAA with acceptable FPS(min 45 avg) in every game I play, I want to run 8xAA across the board. Is ~45FPS avg @ 19x12, 16xAF and 4-8xAA a realistic expectation with a single 4890? Barring games I will never play like Crysis and FC2, despite owning them.

Quote:

Originally Posted by Robilar

Plus it runs my Q9650 at lower vcore than my UD3P requires for stability (don't ask me why, I still can't figure it out).

Kind of tangent to the topic at hand, but a while back when I played around with a UD3P and my sig Q6700, much to my chagrin I noticed the same phenomena with respect to voltage and stability. My tried and true settings of 3.33ghz on the 1066 strap with 1.248 vCORE and 1.24 vNB were not stable with the UD3P, though in all fairness we're talking a measly .02v increase which may be within the margin of error/SW reporting. However, 450FSB which was stable on the LT x48 @ 1.34vNB and 1066 strap needed to be set to ~1.45v on the UD3P.

Long anecdote aside, when going from p45 -> x48 did you notice the same phenomena, where the P45 needs more NB voltage for equal FSB clocks but potentially allows for a higher FSB?

**Robilar** · 04-09-09

Short answer sort of..

Two things to consider (in my case). NB voltage was the same for both boards. The Gigabyte UD3P will do 500 FSB with this chip with minimal tweaks. This X48 board, I cannot get 500 FSB stable at all (yet).

Mind you I don't run the chip at 500 FSB (9x445 is much simpler) so its not a big deal on the Q9650. If I had a chip with a lower multi (thus requiring a higher FSB), then it would be problematic.

I have 2 other X48 boards (rampage formula and a rampage extreme) that I use for other setups but its too much of a hassle to strip them down to test the chip (much as I am curious).

Rochfordrambo · 04-09-09

Please add me

Russtynailz · 04-09-09

Quote:

Originally Posted by BradleyW

how did you get 4890 cf to work. nobody can get cf for those to work. only 1 card is detected for many people

I started off having trouble getting a single card to detect using the 9.2 and 9.3 drivers so i ran the CD that came with the card and BOOM cross fire works.
all my test were done with the factory dirvers. this weekend I plan on installing 9.4 and blowing my scores away!!

no "5.0GHZ I7 LOVE" form me though only 4.4ghz I7 love

(i did move into 4rd place in the Top ATI benchmarks Thread running 2 4890's in cross fire.

third place here I come

pyrophonic · 04-09-09

Quote:

Originally Posted by MasterKromm

I've been intently following this thread, it's been a while since I've upgraded my GPU... While my current GPU will do 19x12, 16xAF and 0-4xAA with acceptable FPS(min 45 avg) in every game I play, I want to run 8xAA across the board. Is ~45FPS avg @ 19x12, 16xAF and 4-8xAA a realistic expectation with a single 4890? Barring games I will never play like Crysis and FC2, despite owning them.

I would love to find the answer of this as well, i have same size monitor and the same expectations of what performance i want out of a card with out spending too much

i play games like tf2, gta IV,fallout 3,stubbs the zombie

,Left 4 dead.

NCspecV81 · 04-09-09

Quote:

Originally Posted by pyrophonic

I would love to find the answer of this as well, i have same size monitor and the same expectations of what performance i want out of a card with out spending too much

i play games like tf2, gta IV,fallout 3,stubbs the zombie

,Left 4 dead.

Check out my thread I do some runs with 8x on other games. I done l4d and averaged 115-118fps @ 1920x1200 8xAA 16xAF.

DeadSkull · 04-09-09

Quote:

Originally Posted by Robilar

I mentioned earlier in this thread that I ran a pair of 4870's on the UD3P without any performance hit. I provided the tweaktown article (and another member referenced the legion article) simply to point out that some tests have shown a difference in FPS in certain games at certain resolutions.

I personally think the tweaktown test was skewed as the results were pretty far off other sites (like Tom's for example). Again I think it may be a case by case basis. The games I run showed no apparent difference in FPS between the P45 and the X48 boards and I stand by that.

for the sake of not completely debunking the posted site reviews about this, perhaps it depends on what res, what game, what ingame settings.. a lot of variables.

The only game that seems to take a noticeable hit at res of 1680x1050 and higher is Crysis.

Quote:

Originally Posted by Robilar

Plus it runs my Q9650 at lower vcore than my UD3P requires for stability (don't ask me why, I still can't figure it out).

Attached is my orthos shot from yesterday's test. Note the temps and vcore and this is before my AS5 had cured.

when I get a chance, I'll move the 4890's over to my other UD3P system and run some benches and compare accordingly.

Why dont you run P95 to test for stability?
Orthos only loads on two cores.