RX 480 vs 1060, several month ago vs now, the AMD driver improving over time is not a myth

[medi01]

I wish TPU would also rerun the tests.

http://www.hardwarecanucks.com/foru.../73945-gtx-1060-vs-rx-480-updated-review.html

TLDR

DX11

On release (July 2016), the GTX1060 was around 12% and 8% better than RX480 in 1080P and 1440P. Now (Dec 2016), the GTX1060 is 2% and 0% better than RX480 in 1080P and 1440P.

DX12

On release (July 2016), the GTX1060 was around 3% and 4% worse than RX480 in 1080P and 1440P. Now (Dec 2016), the GTX1060 is 6% and 6% worse than RX480 in 1080P and 1440P.

 

[TheinsanegamerN]

When TPU benchmarks new games, they use newer drivers. Any improvements AMD makes will be there.

You also have to account for how many games need multiple performance patches to run well. Makes it difficult to see if this is AMD driver improvements or games were truly not coded well to use AMD hardware. Or any hardware, in some game's cases *cough* Dishonored 2 *cough*.

 

[RejZoR]

That's pretty significant really. Now, RX480 takes the lead in pretty much all games. In worst case scenarios it's just the same as GTX 1060.

 

[rtwjunkie]

Or any hardware, in some game's cases *cough* Dishonored 2 *cough*.

I wouldn't say ANY. My system (in system specs) has run it spectacularly well since release day. Honestly, the culprit on NVIDIA side with these new games that seem to have had problems? My anecdotal observation says it boils down to Pascals.

 

[the54thvoid]

Another site looked at 980ti and Fury X.

http://www.babeltechreviews.com/fury-x-vs-gtx-980-ti-revisited-36-games-including-watch-dogs-2/3/

Hope the link works. Fury X gains a little more but the 980ti holds up well enough.

We know AMD improves over time. The short span of the 1060 is worrying although more reviews would be required for proper validation. Given the options, 480 was the better bet even at release, given its better compute chops.

 

[cdawall]

Just give it time the 480 will end up closer to the 1070 than 1060 nvidia is notorious for not caring about the 60 series.

 

[P4-630]

the 480 will end up closer to the 1070

Yeah ofcourse, closer, but that's all, I'm sure it will never reach GTX1070 performance, certainly not at stock speeds.

 

[EarthDog]

MEH, when DX12 is worth something, the RX480 will be worth more... oh wait, their next gen will be out.

Perfectly viable card indeed, more so now in fact. But until I see more DX12 in the market, either card is seemingly a good choice.

Just give it time the 480 will end up closer to the 1070 than 1060 nvidia is notorious for not caring about the 60 series.

closer, sure.. close, lol, NO.

 

[P4-630]

Just give it time the 480 will end up closer to the 1070

 

[cdawall]

Yeah ofcourse, closer, but that's all, I'm sure it will never reach GTX1070 performance, certainly not at stock speeds.

Of course it wont exceed it or even match it for that matter. Even using AMD magical compute its only 5.8 Tflops vs 6.5

 

[P4-630]

Of course it wont exceed it.

I didn't mention exceed either..

 

[cdawall]

I didn't mention exceed either..

Oh settle down I fixed my post clown. For reference the 1060 is 3.8->4.4Tflops of computer performance hence my comment.

 

[P4-630]

Oh settle down I fixed my post clown.

You seem to be the clown here LOL!

 

[cdawall]

You seem to be the clown here LOL!

In what way? The belief that amd can provide a quality product that over time puts them closer to the next model up? That's a common proven thing. 290/x competed with the 970 then the 980 as time progressed fury x the 980 and now the 980ti

 

[EarthDog]

lol, TPU, the site for the blind leading the blind pissing matches!!!!!!!!!!!!!!!

 

[P4-630]

Never been there though...

 

[cdawall]

Never been there though...

Curious I would have expected green doors

 

[HD64G]

A great re-review there indeed. And DX12 shows what deniers don't want to accept. CGN is future proof. Pascal isn't. So, Vega will be the GPU for enthusianst gamers, not any Pascal-based GPU. Let's see when it will launch now. And remember that Vega 10 is the small one and will launch 1st, while Vega 11 is the big core to go against GP100.

 

[phanbuey]

A great re-review there indeed. And DX12 shows what deniers don't want to accept. CGN is future proof. Pascal isn't. So, Vega will be the GPU for enthusianst gamers, not any Pascal-based GPU. Let's see when it will launch now. And remember that Vega 10 is the small one and will launch 1st, while Vega 11 is the big core to go against GP100.

If and when... let's see it- I've been gaming on the 1080 for a few months now and still no Vega; and if it comes out and is around the same speed as a 1080 then current customers have no incentive to upgrade. If it's going to take 6 months to become marginally faster than a 1080, then might as well just wait for the next cycle.

 

[cdawall]

If and when... let's see it- I've been gaming on the 1080 for a few months now and still no Vega; and if it comes out and is around the same speed as a 1080 then current customers have no incentive to upgrade. If it's going to take 6 months to become marginally faster than a 1080, then might as well just wait for the next cycle.

See that's an interesting perspective. I have to say people upgrade for different reasons and a lot of people haven't grabbed a 1080 because they are waiting on the 1080Ti or AMD offerings, most wont wait until the next cycle.

 

[phanbuey]

See that's an interesting perspective. I have to say people upgrade for different reasons and a lot of people haven't grabbed a 1080 because they are waiting on the 1080Ti or AMD offerings, most wont wait until the next cycle.

But you have 2 CF 480's... you're not going to necessarily dump them for a single Vega if the performance is within 5-10% (unless it's to avoid driver issues). The problem with releasing the same performance 6 months later is that you've lost a big chunk of the market that may have otherwise bought in.

I will always buy the fastest I that can afford at the time of upgrade, the fact that it will take a card 6 months to possibly mature in terms of performance, or if I will need to wait 6+ months for a card that is 5-15% faster (or the same) - will always push me to the card that just came out, is available, and is 35-40% faster than anything else.

Point is, having hidden performance in their hardware hurts AMD more than it helps them, because when people are going to buy cards, they are looking at the current benches.

 

[the54thvoid]

A great re-review there indeed. And DX12 shows what deniers don't want to accept. CGN is future proof. Pascal isn't. So, Vega will be the GPU for enthusianst gamers, not any Pascal-based GPU. Let's see when it will launch now. And remember that Vega 10 is the small one and will launch 1st, while Vega 11 is the big core to go against GP100.

Saying 'deniers' makes you sound like an 'apologist'

Anyway, the Vega 10/11 issue is very mixed right now - most sources claim Vega 11 will replace Polaris 10 in the mainstream and Vega 10 is the bigger chip. Also, as much as you tout GCN, I'm pretty sure AMD are dumping/restructuring it for a new architecture? Pretty sure I read that somewhere too?

As for Pascal being not being future proof? You're so very wrong. Pascal has <subject to much confusion> no hardware async (but it does really, just done differently). How can a mid range architecture like GP104 best the gfx charts, even in DX12 AMD titles with Async compute heavy loads? Because it's nuts fast. Then, look at GP102 - it's far ahead of GP104. Full fat Pascal is a freaking monster and it has (in many minds) not very good Async or DX12 chops. So why does it destroy the entire field?

Also, given in some titles a 390X will equal a Fury (non X), it shows GCN is reaching certain limits for compute (probably why vega or navi might be adopting a new approach). Stop banging that AMD drum so hard. And FTR, if you've seen my posts my next build is hopefully Zen/Vega, I am not a brand fan. Though if Vega is late and a 1080ti comes out.....

As for the OP, yeah, 480 is a great card.

 

[cdawall]

But you have 2 CF 480's... you're not going to necessarily dump them for a single Vega if the performance is within 5-10% (unless it's to avoid driver issues). The problem with releasing the same performance 6 months later is that you've lost a big chunk of the market that may have otherwise bought in.

I will always buy the fastest I that can afford at the time of upgrade, the fact that it will take a card 6 months to possibly mature in terms of performance, or if I will need to wait 6+ months for a card that is 5-15% faster (or the same) - will always push me to the card that just came out, is available, and is 35-40% faster than anything else.

Point is, having hidden performance in their hardware hurts AMD more than it helps them, because when people are going to buy cards, they are looking at the current benches.

I only grabbed a second card because it was cheap. I agree people look at current benches to judge card purchases however and the only thing that steered me towards a 480 over the 1060 was xfire ability and some very loud people on this page.

I also have 3 980Ti's in my work machine as well so the 480's are "slower" than that setup, but I personally cannot really see a difference. Nothing short of the Titan X pascal really seem like a worthwhile upgrade for either rig and I hope what AMD drops on the top end competes with that. I know that is a pipe dream, but I can hope.

 

[Mercennarius]

FYI those considering an RX 480 can pick up a Powercolor 4GB 480 for $164.99 with free shipping today:
https://slickdeals.net/f/9498892-powercolor-red-dragon-radeon-rx-480-4gb-for-164-99?v=1

 

[the54thvoid]

I know that is a pipe dream, but I can hope.

If RTG can drop a 1080 matching (in DX11) and 1080 beating (in DX12) Vega 10, the dream is real. Fury X was not too far behind 980ti. Fury X was hampered by the HBM implementation (for voltage lock outs and unknown variables of memory/core overclocks). I don't see why Vega cannot be 50% faster than Fury X. Same number (4096) of cores infers far higher clocks at lower process node. Refined GCN arch would allow Vega to hit the sweet spot though I sincerely doubt it will topple Titan Xp. Only unknown is pricing.... and date.