Thank you! What is a good format for this topic in your opinion? I really want to Livestream for a couple hours, but I have to switch GPUs continuously so people would have to wait 3 minutes for me to come back haha
@@ContradictionDesign It’s a tough call. The livestream is good because people can weigh-in in realtime and make requests but personally I’m a big fan of the shorter/edited videos because I find it easier to retain the information. So no matter what you do, I’m sure it will please at least some people. :)
@@perlichtman1562 That is all true! Thanks for your opinion. That does make sense though. It is also harder to find the important info inside the stream once its posted as a video.
On my 4090 with stock settings, I am getting 34,230 pretty consistently. It's important to turn off Core Isolation (windows feature that can cripple performance).
Great video, just tested my 4090 founders edition and I got 32983 points with almost no background apps running, intel virtualization technology and core isolation are turned off, + im not using studio drivers but gaming ones... how does it sound? all good or something's off? thanks 💕
I still have to check on the core isolation issue. I have never played with that setting. Cinebench varies a ton from run to run, and machine to machine. It is hard to say if your test result is off, because it is hard to standardize a run. I would say try running it again with administrator and see if that helps force it to be the primary application. Also, what is your GPUs power limit? I know these 40 series can get a little bit more performance if they run at max power.
Yes, unfortunately I noticed that this test only tests the Cuda cores, leaving the RT cores idle, and that's actually only half of the video card! They are doing this just when the overall performance of the video cards (Cuda and RT cores) is becoming more and more important! So I won't use this test very much, as I don't have any more gpu systems, it was enough to pay for one :)
ello! Thanks for the new video! I have a question. I wanted to upgrade my computer by buying a new video card, but I have to budget $500. My current system is as follows: i7 11700k, 650 watts (power supply), 32 RAM and a 1060 on 6 gigabyte video card. The purchase of a video card is necessary for computer graphics, that is, in 3D and, less important, for games. So what would you recommend in my case?
For example, I am now thinking about buying an RTX 4060 with 8 gigabytes or Radeon, but I don't know which one. So, if you don't mind, please recommend.
@@hamster8492 I would really try to get a RTX 4070 for around $500. I have seen brand new ones for as low as $530. I know that's over budget but the 4060 ti is not worth saving $50. The 4070 is up to 50% or so faster than the 4060 ti. But the 4070 would get you 12 GB VRAM and very good speed, while being pretty efficient on power.
when the new benchmark software releases that means current GPU is not worth to buy.....I'm so happy because I didn't buy any 40 series GPUs, I hope new GPU will utilize pcie gen 5 and release in the Q1 of the next year.
I have a rtx a4000 the score stock no oveclock is: 10784, but I don't know when start count time, when I click start test or when the render start after preparing project
@@ContradictionDesign thanks for your content, after 30 minute test drop score to 10559, but i forgot to say in the first run the fancurve is 17% more agressive because the a4000 is a single slot and hot, but only 140w, but with stock fan and core the 10 min score drop to 10306, and with fan curve +17% and 150+ overclock the score is only 10827
Very good! I think recording this with GPU encoding took enough power from the benchmark. I ran it right after, and was able to hit 37k with +1000 memory clock and +150 core. Stock settings, I get about 36k. Thanks for sharing your results too!
@@dirtywindshield5034 This benchmark is very sensitive to background apps and system load bottlenecks. So make sure no other apps are open. In the test I recorded for this video, I got thousands of points less than when not recording. And this is using the AV1 encoder, so it should not affect rendering speed much. Other than that, I would have to see your system specs to try to guess what the bottleneck could be
Good result! This test seems very sensitive to background programs. I recorded this on GPU with AV1, and had a browser tab open. Also, I am at stock settings for the GPU power and clocks. I will re-test with OC settings soon.
@@BestFriendOfDog I actually just re ran this with +1000 memory clock, +150 core. Just got it over 36K. It varies a ton without changing settings anyway it seems
@@ContradictionDesign Yeah exactly, i've been extremely busy with work so haven't had enough time to troubleshoot it. I'd like to though for sure, bit concerned if it's the gigabyte board that's the culprit (based on other people's experiences
For context, it got worse after the 23H2 update and resize bar enabling. The score is down to around 7000 currently. EXPO enabled trashed my stability until I disabled it again. Latest drivers installed, BIOS updated. have a 7800X3D on a B650i and a 4090FE
@@NikoBee90 wow that is wild. It sounds like you know what you're doing. It really does sound like the GPU is the issue. It's not overheating at all is it? I have had my 3080 tis from gigabyte leak fluid from their thermal pads, and one does run a bit hotter than the others
@@ContradictionDesign Fraid not, at least from what I can tell so far. Temps always remain low, utilization peaks at 80% or so but it's not steady. Mild coil whine use to happen and now rarely if ever which could be a sign of an internal fault of the card but of something went on it I'd imagine it would be a bit more obvious
Great stuff. Looking forward to more of these. :)
Thank you! What is a good format for this topic in your opinion? I really want to Livestream for a couple hours, but I have to switch GPUs continuously so people would have to wait 3 minutes for me to come back haha
@@ContradictionDesign It’s a tough call. The livestream is good because people can weigh-in in realtime and make requests but personally I’m a big fan of the shorter/edited videos because I find it easier to retain the information. So no matter what you do, I’m sure it will please at least some people. :)
@@perlichtman1562 That is all true! Thanks for your opinion. That does make sense though. It is also harder to find the important info inside the stream once its posted as a video.
On my 4090 with stock settings, I am getting 34,230 pretty consistently. It's important to turn off Core Isolation (windows feature that can cripple performance).
Thank you for the tip! I will look for this setting!
first also awesome video!
Thank you!
Great video, just tested my 4090 founders edition and I got 32983 points with almost no background apps running, intel virtualization technology and core isolation are turned off, + im not using studio drivers but gaming ones... how does it sound? all good or something's off? thanks 💕
I still have to check on the core isolation issue. I have never played with that setting.
Cinebench varies a ton from run to run, and machine to machine. It is hard to say if your test result is off, because it is hard to standardize a run.
I would say try running it again with administrator and see if that helps force it to be the primary application. Also, what is your GPUs power limit? I know these 40 series can get a little bit more performance if they run at max power.
I recommend the studio drivers for this.
I agree. That is normally what I use for my 3d machines
Yes, unfortunately I noticed that this test only tests the Cuda cores, leaving the RT cores idle, and that's actually only half of the video card! They are doing this just when the overall performance of the video cards (Cuda and RT cores) is becoming more and more important! So I won't use this test very much, as I don't have any more gpu systems, it was enough to pay for one :)
Yeah hopefully they'll add an rt test sometime. But it will be interesting to compare shader performance anyways
DAMN that thing doesn't waste any time!!
Yeah it's very fast. Makes blender viewport very useful haha
你的每一篇都有惊喜 真是太棒了！！
Glad you are happy with them! Thanks for watching!
My new rig with the Gigabyte RTX 4090 OC won't pass 30k points...something is definitely not right and I'm not sure what to do
Is the gpu by chance installed on a riser? Another commenter had issues with a riser causing slowdowns
ello! Thanks for the new video! I have a question. I wanted to upgrade my computer by buying a new video card, but I have to budget $500. My current system is as follows: i7 11700k, 650 watts (power supply), 32 RAM and a 1060 on 6 gigabyte video card. The purchase of a video card is necessary for computer graphics, that is, in 3D and, less important, for games. So what would you recommend in my case?
For example, I am now thinking about buying an RTX 4060 with 8 gigabytes or Radeon, but I don't know which one. So, if you don't mind, please recommend.
@@hamster8492 I would really try to get a RTX 4070 for around $500. I have seen brand new ones for as low as $530. I know that's over budget but the 4060 ti is not worth saving $50. The 4070 is up to 50% or so faster than the 4060 ti. But the 4070 would get you 12 GB VRAM and very good speed, while being pretty efficient on power.
@@ContradictionDesign OK, got it, I'll think about it. Thank you!
You are welcome!
@@ContradictionDesign By the way, if you are interested, I work in the following editors: 3D Maya, ZBrusch.
i have a 4090 and wanted to test in cinebench but it isnt letting me run the test.... any idea why?
Well I guess make sure you have the newest drivers and firmware for the GPU, so the test can recognize the device. Do you get error messages from it?
For whatever reason when I got back from vacation it randomly started working! Thx!@@ContradictionDesign
when the new benchmark software releases that means current GPU is not worth to buy.....I'm so happy because I didn't buy any 40 series GPUs, I hope new GPU will utilize pcie gen 5 and release in the Q1 of the next year.
Yes I am looking forward to the future! Should be fun!
22879 points on rtx 4070 super is that good or bad?
I am running a bunch of cinebench tests currently. Your score sounds good but I'll have real numbers for you soon!
@@ContradictionDesign ok thanks man
@@greenlight5009 No problem!
I have a rtx a4000 the score stock no oveclock is: 10784, but I don't know when start count time, when I click start test or when the render start after preparing project
Thanks for sharing your results! The test runs 10 minutes after the scene is prepared. Only thing to report is the score. Thanks for the data!
@@ContradictionDesign thanks for your content, after 30 minute test drop score to 10559, but i forgot to say in the first run the fancurve is 17% more agressive because the a4000 is a single slot and hot, but only 140w, but with stock fan and core the 10 min score drop to 10306, and with fan curve +17% and 150+ overclock the score is only 10827
@@FRBM412 Oh very nice! Thanks! I have wanted to get some a4000 gpus, but just haven't done it yet. They seem like great cards
I scored a 36,260 on my Founders 4090.
Nice! That's a pretty good result! Do you have an overclock on that bad boy?
I just got a 4090 fe i get 35+++ every time
Very good! I think recording this with GPU encoding took enough power from the benchmark. I ran it right after, and was able to hit 37k with +1000 memory clock and +150 core. Stock settings, I get about 36k. Thanks for sharing your results too!
@@ContradictionDesign I have a 4090 FE on a custom water loop. At stock I'm getting 31k what gives?
@@dirtywindshield5034 This benchmark is very sensitive to background apps and system load bottlenecks. So make sure no other apps are open. In the test I recorded for this video, I got thousands of points less than when not recording. And this is using the AV1 encoder, so it should not affect rendering speed much. Other than that, I would have to see your system specs to try to guess what the bottleneck could be
meine schafft out of the box 36000 im cinebench 2024
It is incredibly fast. It is also crazy how much faster this GPU is versus current high end CPUs.
I got 36k with the same 4090 AMP
I had everything closed, had 38k the first time i ran cb24 for some reason... i feel it has quite a bit of variation
Good result! This test seems very sensitive to background programs. I recorded this on GPU with AV1, and had a browser tab open. Also, I am at stock settings for the GPU power and clocks. I will re-test with OC settings soon.
@@BestFriendOfDog I actually just re ran this with +1000 memory clock, +150 core. Just got it over 36K. It varies a ton without changing settings anyway it seems
@@ContradictionDesign I have it at stock too. All I can add is that I am using a 7950x and windows 11 with 6000 cl30 ram
@@ContradictionDesign yeap
Rtx 4080 non super 30078 score
That is actually a great score! I am surprised it is that close to the 4090!
Pff, how about 13,500pts on my 4090...makes ya just wanna cry
13,500? It should run around 35,000. Make sure there are no other processes running on the GPU. This result is very interesting! Want to debug it?
@@ContradictionDesign Yeah exactly, i've been extremely busy with work so haven't had enough time to troubleshoot it. I'd like to though for sure, bit concerned if it's the gigabyte board that's the culprit (based on other people's experiences
For context, it got worse after the 23H2 update and resize bar enabling. The score is down to around 7000 currently. EXPO enabled trashed my stability until I disabled it again.
Latest drivers installed, BIOS updated. have a 7800X3D on a B650i and a 4090FE
@@NikoBee90 wow that is wild. It sounds like you know what you're doing. It really does sound like the GPU is the issue. It's not overheating at all is it? I have had my 3080 tis from gigabyte leak fluid from their thermal pads, and one does run a bit hotter than the others
@@ContradictionDesign Fraid not, at least from what I can tell so far. Temps always remain low, utilization peaks at 80% or so but it's not steady.
Mild coil whine use to happen and now rarely if ever which could be a sign of an internal fault of the card but of something went on it I'd imagine it would be a bit more obvious