Wonder when Nvidia will get their Atomic Heart driver out 😛
The real question is why hell is a game that's been used in Nvidia RT marketing and demos since the 2018 launch of the 2080, not launching with RT lol...
I'm wondering when "High idle power has situationally been observed when using select high-resolution and high refresh rate displays on Radeon RX 7000 series GPUs." will be fixed.
I'm wondering when "High idle power has situationally been observed when using select high-resolution and high refresh rate displays on Radeon RX 7000 series GPUs." will be fixed.
I have a sneaky suspicion that it wont be entirely fixed.
I think the new MCD design is hurting them a little in terms of power draw. The chips on their own are super efficient but combined in a package they are not so much.
We all know moving data consumes a lot of power. This is why memory across the entire PC platform has moved a lot over the years closer and closer next to where its meant to be.
RDNA2 had the IF on the one die, now RDNA3 moves this to their own dies. So I believe (this is just my theory) that consistently moving data between the 4 memory dies too and from memory (VRAM) and too and from the GPU die is causing these high power draws when basically doing nothing or doing simple tasks. Sure the memory is right next to the GPU die but its further away than on RDNA2 and now you have to move the same data twice (VRAM>IF>GPU whereas on RDNA2 it was just VRAM>IF).
I also believe this is why RDNA3 generally hasn't increased core clock speeds over RDNA2 by much, if at all. AMD pushed for an aggressively low energy budget and this I think backfired on them.
They said RDNA3 was desiged to exceed 3GHz, I believe this is AMD being 100% honest and that the GPU core it self can exceed 3GHz easily. But the added energy cost of the MCD design and limiting power target is what is holding it back.
This also ties into them stating recently that RDNA4 will be launching sooner than expected. I think AMD have solved this issue with RDNA4, weather thats through a redesign, node shrink, or basically a higher power target we hopefully will soon find out.
AMD could have easily made RDNA3 as it is a success though by being even more aggressive with pricing. If the 7900XTX was $699-$799 and the XT $599-$699 then I think people would have overlooked these issues.
Oh and release it without a vapour chambor issue.
I have yet to try all my games, but so far it seems more stable than the .1 on my 6800XT
Will see in a few days if it's necessary to roll back one again or not.
I have a sneaky suspicion that it wont be entirely fixed.
I think the new MCD design is hurting them a little in terms of power draw. The chips on their own are super efficient but combined in a package they are not so much.
We all know moving data consumes a lot of power. This is why memory across the entire PC platform has moved a lot over the years closer and closer next to where its meant to be.
RDNA2 had the IF on the one die, now RDNA3 moves this to their own dies. So I believe (this is just my theory) that consistently moving data between the 4 memory dies too and from memory (VRAM) and too and from the GPU die is causing these high power draws when basically doing nothing or doing simple tasks. Sure the memory is right next to the GPU die but its further away than on RDNA2 and now you have to move the same data twice (VRAM>IF>GPU whereas on RDNA2 it was just VRAM>IF).
I also believe this is why RDNA3 generally hasn't increased core clock speeds over RDNA2 by much, if at all. AMD pushed for an aggressively low energy budget and this I think backfired on them.
They said RDNA3 was desiged to exceed 3GHz, I believe this is AMD being 100% honest and that the GPU core it self can exceed 3GHz easily. But the added energy cost of the MCD design and limiting power target is what is holding it back.
This also ties into them stating recently that RDNA4 will be launching sooner than expected. I think AMD have solved this issue with RDNA4, weather thats through a redesign, node shrink, or basically a higher power target we hopefully will soon find out.
AMD could have easily made RDNA3 as it is a success though by being even more aggressive with pricing. If the 7900XTX was $699-$799 and the XT $599-$699 then I think people would have overlooked these issues.
Oh and release it without a vapour chambor issue.
That is my suspicion as well I'm afraid. The architecture design is to blame.
Same as last driver, Eyefinity is broken. You get 1 monitor that works and black screen on the others. 6800XT has had bad drivers for a while now. Random black screens and tearing are driving me nuts.
I'm wondering when "High idle power has situationally been observed when using select high-resolution and high refresh rate displays on Radeon RX 7000 series GPUs." will be fixed.
Same as last driver, Eyefinity is broken. You get 1 monitor that works and black screen on the others. 6800XT has had bad drivers for a while now. Random black screens and tearing are driving me nuts.
You have had bad driver issues, but you can not speak for everyone and say 6800xt has had bad drivers. Thats an ignorant blanket statement.
Your personal issues is not all be all issues.
Guess what, 0 issues here with my 6800xt on my superwide, and all eyefinity with 2 27 inch side monitors. In fact, 2015 CCC options for eyefinity hotkey still works,
I'm wondering when "High idle power has situationally been observed when using select high-resolution and high refresh rate displays on Radeon RX 7000 series GPUs." will be fixed.
this has been a problem for like a decade, its low priority since it only effects a small subset of monitors, and its not enitrely amd's fault. I dont expect it to get fixed any time soon,
some monitor manufacurers set their vesa timings too tight , and there isnt enough time for the gpu to switch power states during the vblank period. if they were to switch outside of vblank it results in flickering, so the dirty fix is to keep the idle state always high...
I have yet to try all my games, but so far it seems more stable than the .1 on my 6800XT
Will see in a few days if it's necessary to roll back one again or not.
May I respectfully inquire what instabilities you experienced with 23.2.1?
***
I always know when AMD will release a new driver; it will be within 36 hours of myself installing the driver it superseded.
We all know moving data consumes a lot of power. This is why memory across the entire PC platform has moved a lot over the years closer and closer next to where its meant to be
More specifically the actual computation is almost free ! The energy that is consumed to do the calculation is nothing but a tiny fraction of what you spend to move the data for the calculation! That said by the same logic the mcm design approach can also save joules as far the data in there is the most used ones hundreds or thousands of times ... It will save you lots of joules! In theory, but if there is not enough space or data that is clearly used a lot more often than other pieces, then potentially this can boomerang and spend more joules! At least in theory people can correct me if I am wrong I am guessing mostly based on the admittedly small knowledge I have on the matter !
More specifically the actual computation is almost free ! The energy that is consumed to do the calculation is nothing but a tiny fraction of what you spend to move the data for the calculation! That said by the same logic the mcm design approach can also save joules as far the data in there is the most used ones hundreds or thousands of times ... It will save you lots of joules! In theory, but if there is not enough space or data that is clearly used a lot more often than other pieces, then potentially this can boomerang and spend more joules! At least in theory people can correct me if I am wrong I am guessing mostly based on the admittedly small knowledge I have on the matter !
I just hope that they will learn how to acquire the benefits of the power design with the 7900 series, so they can mitigate this issue or at least have a small workaround. Perhaps that's too much to wish for, but yeah. I wasn't planning on not using my monitors to their full capabilities. My RTX 3080 powered these monitors just fine, so it feels weird that they're suddenly problematic. It's just a little weird, because they are FreeSync monitors. I have all the patience in the world, I'll wait if it takes a year.
- out of the blue stutters in some of the games I play (which were not there in 22.11.1)
- random graphical glitches in No Man's Sky, Satisfactory and Subnautica (again, not experienced in 22.11.1)
- completely random Grim Dawn crashes (which never happened before, not on nVidia, not on AMD before this driver)
- completely sudden playback error while playing Youtube videos (had to restart browser, happened multiple times)
And stuff was somehow "off" with it, like I had lag when I shouldn't, or very long game loading times.
My guess is some kind of shader compilation/execution bug got into it that affects certain games with older engines (or all of them, I don't know, I don't have the latest just released games, I usually don't jump on them as soon as they are launched)
However, so far in this 2.2 I don't seem to experience any of these today and yesterday, so I guess they just found what was wrong a few days after releasing that driver which resulted in the very quick release of this new version.
You have had bad driver issues, but you can not speak for everyone and say 6800xt has had bad drivers. Thats an ignorant blanket statement.
Your personal issues is not all be all issues.
Guess what, 0 issues here with my 6800xt on my superwide, and all eyefinity with 2 27 inch side monitors. In fact, 2015 CCC options for eyefinity hotkey still works,
You lie, you know why I say that, I run MST sim center in the USA, everyone of our systems run triples with AMD 6000 & 7000 gpus. No issues with 7000 gpu's. But the 17 6800XT systems are down because Eyefinity is broken. Don't bullshit people young man.
You lie, you know why I say that, I run MST sim center in the USA, everyone of our systems run triples with AMD 6000 & 7000 gpus. No issues with 7000 gpu's. But the 17 6800XT systems are down because Eyefinity is broken. Don't bullshit people young man.
Did you try 22.Q4? For me I'm stick with pro driver 22.Q4 since I have random freeze with the driver after 22.11.1. I have no idea what the cause of it sometime during work in photoshop some time on gaming.
I have 2 PC with 6700XT while other 2 with RX6600 they have same problem not sure they fix the problem or not.
Also I have black screen during install report to them many time. Now I have to use only device manager to install driver to avoid black screen during install / it happens to random pc with windows 10/11 even fresh install
But so far 22.Q4 is running ok for me no black screen or freeze during work. All 4 pc no problem with 22.Q4 just one black screen during uninstall driver with AMD setup. But so far happy 🙂
I just hope that they will learn how to acquire the benefits of the power design with the 7900 series, so they can mitigate this issue or at least have a small workaround. Perhaps that's too much to wish for, but yeah. I wasn't planning on not using my monitors to their full capabilities. My RTX 3080 powered these monitors just fine, so it feels weird that they're suddenly problematic. It's just a little weird, because they are FreeSync monitors. I have all the patience in the world, I'll wait if it takes a year.
Thanks Venix. I really like this GPU a lot and it is very powerful, the cooler on the thing is magnificent and really solid. Fan speed does not get near 50% under heavy load playing Dying Light 2 at 1440p Ultra Quality with Raytracing and temperatures are normal. I love that game.
I want both AMD and NVIDIA to be victorious, so that we consumers get the benefits. These driver problems are not helping AMD, because they actually have a good product, but because of these drivers it looks bad. I don't care how long it takes for them to fix it, I just want them to have a solid product. The problem being in the "Known issues" section of the drivers gives me a little hope that they will fix it in the end. My second monitor is not doing much anyway, except giving me more room to stuff app windows in. I've been on the red team and green team endless times, so GPU problems are not new. I remember the GTX 670 FTW needed a BIOS update to fix stuttering, so yeah. Been there, done that.
***
I always know when AMD will release a new driver; it will be within 36 hours of myself installing the driver it superseded.