Become a MacRumors Supporter for $25/year with no ads, private forums, and more!

MP 1,1-5,1 GTX 680 in a 3,1 x Catalina Performance "quirk"

z33capt13n

macrumors newbie
Original poster
Apr 1, 2020
17
7
So I've had a browse in the forums but I've not come across this particular quandry.

I've got a 2008, 8 core (3,1) mac pro, Catalina installed with DosDude1's patcher. I installed and flashed a windows EVGA Nvidea GTX 680 (2gb) card. Functionally it's great, even has boot screen. HOWEVER I've been chatting to a pal and we've compared benchmarks and in both Unigen Valley and Geekbench, my GPU is falling way behind where it should be performance wise. He has a GTX 750 ti in his 3,1 and he's seeing similar scores to me. Even with metal. According to what we could dig up, my card should be way way faster.

Figures I found online put it at 10725 for metal computing but my actual score on Geekbench 5 was 8113
And It scored 1423 in unigen valley, High settings, no AA and 1650x1050 fullscreen (the settings said pal used).


Anyone know what could be causing it?
My guess was either the patcher or the lack of catalina drivers for an old-ass card.

I guess its not a huge issue as this machine is what it is (a 12 year old machine which has been dragged by the scruff of its neck into 2020 with some software and hardware trickery) and it works just fine for the design work and general pratting about I bought it for, I've even played some games on it (mad max and Dawn of war 3- way more modern titles than I thought this machine could handle). I'm just curious if anyone else has experienced this weirdness?
 
  • Wow
Reactions: r6mile

h9826790

macrumors G5
Apr 3, 2014
14,716
7,165
Hong Kong
That's pointless. There is no driver for 750Ti in Catalina, which means he ran the benchmark on another platform.

If you want to compare the GPU, you should use the same computer, same OS, same driver, same build of the benchmarking software, but only difference GPU.

GB5 also not a good benchmarking software. The scores vary a lot even on the same computer.

If the GB5 is something like 50% lower than it should, then it can be a sign that something is wrong. But if 10-20% lower than what you find on the net, that most likely means nothing.

For compute performance, it's way easier to compare that in Luxmark. Of course, you should still only compare to other 3,1 which is running Catalina. If within 10% of what you find (not the highest score, but average), then I will say nothing to worry about.

For Unigine Valley, you guys should use the most demanding preset. Otherwise, again, the results may not comparable. For Unigine Valley, which means, the Extreme HD preset.
 
  • Like
Reactions: z33capt13n
Comment

z33capt13n

macrumors newbie
Original poster
Apr 1, 2020
17
7
That's pointless. There is no driver for 750Ti in Catalina, which means he ran the benchmark on another platform.

If you want to compare the GPU, you should use the same computer, same OS, same driver, same build of the benchmarking software, but only difference GPU.

GB5 also not a good benchmarking software. The scores vary a lot even on the same computer.

If the GB5 is something like 50% lower than it should, then it can be a sign that something is wrong. But if 10-20% lower than what you find on the net, that most likely means nothing.

For compute performance, it's way easier to compare that in Luxmark. Of course, you should still only compare to other 3,1 which is running Catalina. If within 10% of what you find (not the highest score, but average), then I will say nothing to worry about.

For Unigine Valley, you guys should use the most demanding preset. Otherwise, again, the results may not comparable. For Unigine Valley, which means, the Extreme HD preset.
Well, that's pretty comprehensive :) thanks for taking the time to reply. I'll ignore the numbers then!
 
  • Like
Reactions: h9826790
Comment
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.