Nvidia vs AMD

[Prysin]

No. I do know that matching memory speeds doesn't help. Ether raising the 290 or limiting the 390. And last night in Fallout I went over the 4g that should be the cap because of the 290. I haven't found anyone else mixing the 200 and 300 series ether. I also have to run the 300 series driver on the 290. Might try running the 200 series driver on the 390 and see if that helps.

hmm.... i would sell the 290, add in some funds and get a second 390.

 

or sell both and get a Fury... remember that while it "only" has 4GB of HBM, the bandwidth effectively make it 5-6GB worth of buffer, as it can move so massive amounts of data back and forth.

 

also, Sapphire Fury looks ballin

[Prysin]

FPS per dollar? R9 290 is the king here I believe

nope...

 

R9 295x2 at a ballin 50-75% discount is...

i paid 550 USD for mine, normal retail is between 1500 and 1000 USD

[WangKe]

I have to admit I had The Witcher 3 as soon as it released too, but that's cause I got a key off a friend for free (he bought a GTX 960 and the store accidentally sent him two so he gave one to me ) I was running a GTX 660 back then, so the performance was a little shit cause Kepler GPUs sucked in TW3 in the early stages, a 780Ti was getting beaten by a GTX 960...

 

 

A retail price for MSI Gaming 780Ti was 710$ when it came out 780Ti was losing to a GTX 960 in The Witcher 3 when it was released cause Kepler got no driver support for that game when it came out for some reason

Well, after a christmas sale and the store owner was my uncle, massive discount. But yeah, around 2/3rd's of a grand's worth of a GPU... eaten by a card half that price, even losing to a card thats cheaper than that.

I want to backhand every person in Nvidia Corp, or shit on their office... literally.

[Morgan MLGman]

What fps were you getting with a 660?

With the raw game, before any patches came out, around 40 FPS average with dips to sub 30 and spikes to 55, mostly within the 35-45 range. It was a card from Gainward (I know, ugh, I wasn't the one choosing it), a Golden Sample one, which means basically with a slight factory OC. On a 1920x1200 resolution, Low-Med settings.

[Morgan MLGman]

nope...

 

R9 295x2 at a ballin 50-75% discount is...

i paid 550 USD for mine, normal retail is between 1500 and 1000 USD

Well, a good discount is always a good discount ^^ I got my MSI Gaming R9 290X for 2/3rd the price of a GTX 970 and it's faster than a 970 so yeah... I feel you

I'm talking on average, you can get a 290 for 199$ without any discounts, considering it's only a little behind a 390 in performance, I'd say it's a pretty good deal

[App4that]

i too do, as this shouldnt happen. the GPUs are by all means the same. and CF is known to work, regardless of RAM capacity differences and clock speed differences. As long as the GPUs are the same GCN version, they should CF.

This simply doesnt seem normal.

The best I have found so far is Skyrim. If I set it to 1440 I get a constant 58fps. In both crossfire using the Skyrim profile and just the 390. Literally the best I've found is being the same :lol:

 

hmm.... i would sell the 290, add in some funds and get a second 390.

 

or sell both and get a Fury... remember that while it "only" has 4GB of HBM, the bandwidth effectively make it 5-6GB worth of buffer, as it can move so massive amounts of data back and forth.

 

also, Sapphire Fury looks ballin

I'm pretty attached to my Vapor-x, can't help it, it's damn sexy. If I had to sell both the temptation to try Nvidia would be real, the reason I posted here. For what ever reason I have horrible luck in my choice in games, but I shouldn't suffer because Nvidia has a better relationship with developers. I've seen some improvements in the UI, but the struggle is real.

[don_svetlio]

With the raw game, before any patches came out, around 40 FPS average with dips to sub 30 and spikes to 55, mostly within the 35-45 range. It was a card from Gainward (I know, ugh, I wasn't the one choosing it), a Golden Sample one, which means basically with a slight factory OC. On a 1920x1200 resolution, Low-Med settings.

Oh wow, so my 750 Ti (860M) is as good as a 660 in Witcher 3 - loooool (sorry, i am happy I got a decent lappy )

[don_svetlio]

The best I have found so far is Skyrim. If I set it to 1440 I get a constant 58fps. In both crossfire using the Skyrim profile and just the 390. Literally the best I've found is being the same :lol:

 

I'm pretty attached to my Vapor-x, can't help it, it's damn sexy. If I had to sell both the temptation to try Nvidia would be real, the reason I posted here. For what ever reason I have horrible luck in my choice in games, but I shouldn't suffer because Nvidia has a better relationship with developers. I've seen some improvements in the UI, but the struggle is real.

Don't rush - 60% of DX12 titles announced are partnered with AMD

[WangKe]

Don't rush - 60% of DX12 titles announced are partnered with AMD

Did that 60% come from devs and publishers that saw what happened in Ashes of Singularity?

[don_svetlio]

Did that 60% come from devs and publishers that saw what happened in Ashes of Singularity?

Probably

[Prysin]

Vapor-X have entirely Sapphire-designed PCBs while Nitro are reference - can that affect it?

could be, impossible to tell unless you have a plethora of cards to test against eachother...

 

you would need a custom PCB 390, 2x reference PCB 390, a custom PCB 290 and a reference PCB 290....

then you also need a reference PCB 290 with 8GB VRAM and a custom PCB 290 with 8GB VRAM to see if the VRAM discrepancy is the issue.

[App4that]

Don't rush - 60% of DX12 titles announced are partnered with AMD

That's great news for single card performance, I've yet to see Radeon address the crossfire issues. Which admittedly could be mine because of the generation difference between the two cards. If that was the case you'd think it would be something Radeon would warn of, or address with software.

[Morgan MLGman]

Oh wow, so my 750 Ti (860M) is as good as a 660 in Witcher 3 - loooool (sorry, i am happy I got a decent lappy )

haha, good for you, I have no idea why the GTX 660 didn't want to overclock, like at all... regardless, perhaps the performance would be a little better now after patches and driver updates so it would edge out the 750Ti, but I'm not certain. It should be faster than a 750Ti, but then again, Kepler vs Maxwell...

[WangKe]

Probably

The feelings I have for Nvidia are the same that I have with WWE. I hope the competition fucks them up real hard.

[don_svetlio]

haha, good for you, I have no idea why the GTX 660 didn't want to overclock, like at all... regardless, perhaps the performance would be a little better now after patches and driver updates so it would edge out the 750Ti, but I'm not certain. It should be faster than a 750Ti, but then again, Kepler vs Maxwell...

I do get the absolute max boost clock since mine is a cool-ass card - talking 1100core when it should be 1000 from the Lenovo spec sheet - 960M is the 1100 860M usually lol

 

That's great news for single card performance, I've yet to see Radeon address the crossfire issues. Which admittedly could be mine because of the generation difference between the two cards. If that was the case you'd think it would be something Radeon would warn of, or address with software.

Impossible to tell

[Prysin]

That's great news for single card performance, I've yet to see Radeon address the crossfire issues. Which admittedly could be mine because of the generation difference between the two cards. If that was the case you'd think it would be something Radeon would warn of, or address with software.

i think at this point, we should create a separate thread on this topic, to avoid derailing this thread too far...

 

we already took up 3 pages

[don_svetlio]

could be, impossible to tell unless you have a plethora of cards to test against eachother...

 

you would need a custom PCB 390, 2x reference PCB 390, a custom PCB 290 and a reference PCB 290....

then you also need a reference PCB 290 with 8GB VRAM and a custom PCB 290 with 8GB VRAM to see if the VRAM discrepancy is the issue.

@LinusTech ?

[don_svetlio]

i think at this point, we should create a separate thread on this topic, to avoid derailing this thread too far...

 

we already took up 3 pages

ALL THE DERAILMENT!

[Morgan MLGman]

I do get the absolute max boost clock since mine is a cool-ass card - talking 1100core when it should be 1000 from the Lenovo spec sheet - 960M is the 1100 860M usually lol

I mean... What? How? What are the temps? I've heard about the Y50-70's getting pretty toasty under load...

[WangKe]

ALL THE DERAILMENT!

Not derailed enough

[don_svetlio]

I mean... What? How? What are the temps? I've heard about the Y50-70's getting pretty toasty under load...

68*C GPU max - average 65*C

75*C CPU max - average around 70-72*C

That's when playing CS:GO without V-sync - GPU is maxed, CPU at 50%

EDIT: My friend has the same Y50 - his CPU idles at 45-48. Mine idles at 38-40

my GPU is 86% ASIC - his is 65%

[Prysin]

@LinusTech ?

whilst @LinusTech could probably help, this would be more of a job for PCperspective since that seems to be sort or the things they do..... Going deep down into problems that is very specific in a very specific use case isnt the direction LMG is heading these days...

 

he could probably do it, but AFAIK, he doesnt have the cards needed, and the time it would take to source all these/buy them wouldnt be worth the hassle to just adress a single members issue

 

would make for an interesting case study if nothing else. But not worth it for 90%+ of his audience

[don_svetlio]

whilst @LinusTech could probably help, this would be more of a job for PCperspective. Going deep down into problems that is very specific in a very specific use case isnt the direction LMG is heading these days...

 

he could probably do it, but AFAIK, he doesnt have the cards needed, and the time it would take to source all these/buy them wouldnt be worth the hassle to just adress a single members issue

 

would make for an interesting case study if nothing else. But not worth it for 90%+ of his audience

I'm sad really - 2 years ago Linus would be all over this idea. Now he wouldn't even consider it

[Morgan MLGman]

68*C GPU max - average 65*C

75*C CPU max - average around 70-72*C

That's when playing CS:GO without V-sync - GPU is maxed, CPU at 50%

EDIT: My friend has the same Y50 - his CPU idles at 45-48. Mine idles at 38-40

my GPU is 86% ASIC - his is 65%

It's possible you got very, very lucky ^^ I'm happy for you

[Prysin]

I'm sad really - 2 years ago Linus would be all over this idea. Now he wouldn't even consider it

2 years ago, he didnt have 10-15 employees under him that had bills to pay