FuryX for $519cdn

MasterGoa said:
So you are totally OK with nvidia rendering incomplete frames missing the required textures?
Click to expand...

It's clear you don't understand what I'm saying, and you're not going to open your eyes, so this will be my last post in this thread.

The GTX970 4GB (aka 3.5GB because NV are scummy ****s) does NOT have enough vRAM to load the HQ textures, so it uses a lower-quality stand-in until it can swap the texture from the HDD. Swapping from the HDD is slow as ****, so you are able to run up to a wall and literally watch the HQ textures be swapped in. If you ran DOOM2016 on an NV card that had the same amount of vRAM as the RX480 8GB that he used in the tests, you wouldn't see this. It's a direct symptom of not having enough vRAM. It also doesn't help that the GTX970 only has 3.5GB of actual GDDR5 vRAM and the other .5GB is some slow garbage that shouldn't even count. The more vRAM, the less swapping you would see. Most engines now are caching into the vRAM so it'll utilize more and more if there's a larger pool available. This has been a thing since the beginnings of the Frostbite engine, or at least that's when I remember people talking about it.

If you overload the vRAM, it's going to have **** quality textures until the swap happens, because swapping from the HDD is so much slower than the cached textures from vRAM. This happens with both AMD and NV. This is not new..
 
Nunz said:
It's clear you don't understand what I'm saying, and you're not going to open your eyes, so this will be my last post in this thread.

The GTX970 4GB (aka 3.5GB because NV are scummy ****s) does NOT have enough vRAM to load the HQ textures, so it uses a lower-quality stand-in until it can swap the texture from the HDD. Swapping from the HDD is slow as ****, so you are able to run up to a wall and literally watch the HQ textures be swapped in. If you ran DOOM2016 on an NV card that had the same amount of vRAM as the RX480 8GB that he used in the tests, you wouldn't see this. It's a direct symptom of not having enough vRAM. It also doesn't help that the GTX970 only has 3.5GB of actual GDDR5 vRAM and the other .5GB is some slow garbage that shouldn't even count. The more vRAM, the less swapping you would see. Most engines now are caching into the vRAM so it'll utilize more and more if there's a larger pool available. This has been a thing since the beginnings of the Frostbite engine, or at least that's when I remember people talking about it.

If you overload the vRAM, it's going to have **** quality textures until the swap happens, because swapping from the HDD is so much slower than the cached textures from vRAM. This happens with both AMD and NV. This is not new..
Click to expand...
sorry not seeing it at 4k on 4gb fury x cfx

maybe the 970 just sucks
 
Nunz said:
It's clear you don't understand what I'm saying, and you're not going to open your eyes, so this will be my last post in this thread.
Click to expand...

I do understand what you are saying quite well, but it seems that you dont.
You are OK with nvidia rendering a frame that does not correspond to the selected quality level
so as to maintain a framerate. Telling other they do not "understand" is not a way to validate an erroneous situation...

If high quality textures were selected, the driver should wait for them to load and slow framerate accordingly.
This is not what happenes as per your explanation. So yes, nvidia is using the wrong
textures to maintain framerate. Hence cheating for FPS in benchies...
 
After playing with for a day or so now, ill admit im quite happy with it. Idles at 27c(15% fan) max load temp is around 51c(19% fan) And zero noise. Rig sounds the same at a idle as full load. God i dont miss the old blowers lol. Performance is inline with my expectatations. Games that had good crossfire profiles are a little slower compared to my 290cfx setup.(dying light for example) Games that had zero or poor cfx support run alot better(witcher3)
 
Higgy10 said:
After playing with for a day or so now, ill admit im quite happy with it. Idles at 27c(15% fan) max load temp is around 51c(19% fan) And zero noise. Rig sounds the same at a idle as full load. God i dont miss the old blowers lol. Performance is inline with my expectatations. Games that had good crossfire profiles are a little slower compared to my 290cfx setup.(dying light for example) Games that had zero or poor cfx support run alot better(witcher3)
Click to expand...

Awesome. Glad to hear it's running well. You don't even need to worry about setting blower fan speeds since the FuryX has no blower at all. Very nice. That's the only downside to the hybrid coolers on the NV side -- if you're pumping extra voltage/high clocks, you still need to make sure the blower runs at a moderate speed to keep some cool air over the VRMs, otherwise you'll just bake them.

I'm still surprised the HAF932 required the 240mm fan to come off. I thought for sure you could mount it with the 240mm still on, like you can with the HAF-X. Maybe that's the difference between the two cases, because otherwise they are very similar.
 
Nunz said:
Awesome. Glad to hear it's running well. You don't even need to worry about setting blower fan speeds since the FuryX has no blower at all. Very nice. That's the only downside to the hybrid coolers on the NV side -- if you're pumping extra voltage/high clocks, you still need to make sure the blower runs at a moderate speed to keep some cool air over the VRMs, otherwise you'll just bake them.

I'm still surprised the HAF932 required the 240mm fan to come off. I thought for sure you could mount it with the 240mm still on, like you can with the HAF-X. Maybe that's the difference between the two cases, because otherwise they are very similar.
Click to expand...

Would have to hack something together, only 2 screws on the rad would have lined up so i didnt even bother. Tbh doesnt seem to make a difference on temps but my cpu and gpu are both under water now.

Also despite the good cooler, little reason to pump any extra volts in it lol
 
Higgy10 said:
After playing with for a day or so now, ill admit im quite happy with it. Idles at 27c(15% fan) max load temp is around 51c(19% fan) And zero noise. Rig sounds the same at a idle as full load. God i dont miss the old blowers lol. Performance is inline with my expectatations. Games that had good crossfire profiles are a little slower compared to my 290cfx setup.(dying light for example) Games that had zero or poor cfx support run alot better(witcher3)
Click to expand...

Max load 51c? Damn, that is really good.

I'm pushing a bit extra voltage into mine and OCing to 1100... I also adjusted target temp to 65c (down from default 75c). It runs at about 63c with fan in 30-40% range.
 
SoCalMX70 said:
Max load 51c? Damn, that is really good.

I'm pushing a bit extra voltage into mine and OCing to 1100... I also adjusted target temp to 65c (down from default 75c). It runs at about 63c with fan in 30-40% range.
Click to expand...

ya i just left the tomb raider bench running for 20 odd min and thats top temp ive seen so far
 
Color me green. (not money, envy) But I can bounce back when I replace my R290 with a Vega card next year after some other upgrades.
 
You must log in or register to reply here.
Back
Top