Now We All See the Genius of AMD Going Lowend First

[raghu78]

I like how, in the 1060 review, the 1060 was 7% faster than the reference 480 and now two days later in this review the 1060 is 14.4% faster than the ref 480 and 11% faster than the strix.

note that there is one game missing in this review, WoW which was previously in favor of the 1060.

1060 review (July 19th):

Strix review (July 21st):

there is definitely a mistake as tpu removed Wow which was a strong game for GTX 1060. The perf gap should be smaller and not grow. wizzard said he is looking into it.

https://www.techpowerup.com/forums/threads/asus-rx-480-strix-8-gb.224311/page-2#post-3492873

https://www.techpowerup.com/forums/threads/asus-rx-480-strix-8-gb.224311/page-2#post-3492866

wizzard confirmed there has been an error even with original gtx 1060 review. The GTX 1060 is 11% faster than 480. thats understandable since the only DX12 game they test is RoTR which runs better on Nvidia GPUs. Hitman is still DX11 even though there is a DX12 version which improves performance significantly for AMD GPUs. I think suffice to say that as more DX12 / Vulkan games are added the Rx 480 starts to look better.

 

[USER8000]

there is definitely a mistake as tpu removed Wow which was a strong game for GTX 1060. The perf gap should be smaller and not grow. wizzard said he is looking into it.

https://www.techpowerup.com/forums/threads/asus-rx-480-strix-8-gb.224311/page-2#post-3492873

https://www.techpowerup.com/forums/threads/asus-rx-480-strix-8-gb.224311/page-2#post-3492866

If you add the aggregate FPS its around 5.65% in favour of the GTX1060,but apparently he weights the score and now it is 8% - he also tested the card in the standard mode,which meant it ran at 1291MHZ. Computerbase.de tested all modes.

 

[Silverforce11]

If you add the aggregate FPS its around 5.65% in favour of the GTX1060,but apparently he weights the score and now it is 8% - he also tested the card in the standard mode,which meant it ran at 1291MHZ. Computerbase.de tested all modes.

Did not know he does manipulation of the results based on a personal "weighting" system.. that's messed up and should have been disclosed properly, way earlier.

/facepalm

What's so hard about adding up the numbers, averaging the results. Done and dusted. Nobody could argue that approach is biased.

 

[Thala]

Did not know he does manipulation of the results based on a personal "weighting" system.. that's messed up and should have been disclosed properly, way earlier.

It is not weighted. I gave an example how TPU is calculating the relative score here and why he is doing it still wrong.

What's so hard about adding up the numbers, averaging the results.

Because it skews the result towards games with higher fps if you just add up fps numbers.

 

[raghu78]

If you add the aggregate FPS its around 5.65% in favour of the GTX1060,but apparently he weights the score and now it is 8% - he also tested the card in the standard mode,which meant it ran at 1291MHZ. Computerbase.de tested all modes.

No wizzard is doing it correctly. he calculates each game perf for other cards relative to card being tested kept at 100%. Then he averages the score across all games. Its mathematically the correct thing to do. This way high fps games do not affect the average more than low fps games.

 

[SolMiester]

I'm not sure if you're joking or not.

Anyway,

First chart:
1060=100
480=93
ratio: 100/93=7.5%

Second chart:
1060=108
480=97
108/97 = 11.3%

Note that they've already admitted and fixed their problem in the second chart, they changed the image from what it was when I first posted. The numbers are a bit different now.

Apparently they've been doing wrong calculations for a while now, although as already stated in another thread- they've got other problems with their math.

you are still using different reference points....1060 is 100%, you need to use same for second chart, then show ASUS against that.
Its using ASUS as reference so everything has diff % against that.

 

[USER8000]

No wizzard is doing it correctly. he calculates each game perf for other cards relative to card being tested kept at 100%. Then he averages the score across all games. Its mathematically the correct thing to do. This way high fps games do not affect the average more than low fps games.

It is not weighted. I gave an example how TPU is calculating the relative score here and why he is doing it still wrong.



Because it skews the result towards games with higher fps if you just add up fps numbers.

Saw this in the thread on their forums:

https://www.techpowerup.com/forums/...84U.png&hash=fc3c60ad6a29989cbbfb6fd616c3e0ea

The games with the largest victory are for the GTX1060,yet him trying to "not skew" the results makes the score for the GTX1060 look better than if you just looked at the aggregate FPS. Three of the titles tested had more than 20% wins for the GTX1060 and only one title tested had anything near to a 20% win for the RX480. There are far more >10FPS victories for the GTX1060 in his review over the RX480 than vice-versa,so an aggregate score would favour the GTX1060.

:thumbsup:

 

[Zstream]

Are you serious? Look at the card. Except for the GPU and RAM they are completely different. Nothing's the same.

No, I'm serious. It wasn't meant in jest. I haven't compared pcb boards on either, but did notice an 8pin vs. 6pin. I'll be taking my reference 480 apart here soon, at least when usps delivers it.

 

[railven]

Saw this in the thread on their forums:

https://www.techpowerup.com/forums/...84U.png&hash=fc3c60ad6a29989cbbfb6fd616c3e0ea

<snip>

The games with the largest victory are for the GTX1060,yet him trying to "not skew" the results makes the score for the GTX1060 look better than if you just looked at the aggregate FPS. Three of the titles tested had more than 20% wins for the GTX1060 and only one title tested had anything near to a 20% win for the RX480. There are far more >10FPS victories for the GTX1060 in his review over the RX480 than vice-versa,so an aggregate score would favour the GTX1060.

:thumbsup:

It gets even more in favor of NV's hardware when you benchmark MMOs and some Japanese ports.

I doubt those titles will jump on the DX12/Vulkan bandwagon for a long time.

 

[frozentundra123456]

Saw this in the thread on their forums:

https://www.techpowerup.com/forums/...84U.png&hash=fc3c60ad6a29989cbbfb6fd616c3e0ea

The games with the largest victory are for the GTX1060,yet him trying to "not skew" the results makes the score for the GTX1060 look better than if you just looked at the aggregate FPS. Three of the titles tested had more than 20% wins for the GTX1060 and only one title tested had anything near to a 20% win for the RX480. There are far more >10FPS victories for the GTX1060 in his review over the RX480 than vice-versa,so an aggregate score would favour the GTX1060.

:thumbsup:

One can argue all day about how the calculate the results, but bottom line is that the performance is pretty much a draw. I think one would have a hard time telling the difference in the gameplay in nearly any of the games if he were playing blinded to which card was in use. 1060 does have an edge in the 3 lowest performing games where the games are borderline playable, but more playable on the 1060.

Basically though it comes down to price, and whether better efficiency and a slight edge in most current games is equally or more important than the possible better aging of GCN. So after hundreds of posts and partisan arguing in multiple threads, it is basically a toss up.

 

[sirmo]

Yup 1060 - rx480 are about equal in performance today (AIB vs AIB). Certanly not a noticeable difference anyone would notice in most games.

rx480 pros:
- more VRAM
- FreeSync support (more choice and cheaper)
- crossfire support on the off chance you might want to get a second one at some point
- should perform better with most future titles
- slightly cheaper
- decent crypto miner if that's your thing

1060 pros:
- better legacy game performance, so if you play old games or a specific gameworks title a lot it might be a better choice
- 30 less watts of power used on average..

 

[Silverforce11]

One can argue all day about how the calculate the results, but bottom line is that the performance is pretty much a draw. I think one would have a hard time telling the difference in the gameplay in nearly any of the games if he were playing blinded to which card was in use. 1060 does have an edge in the 3 lowest performing games where the games are borderline playable, but more playable on the 1060.

Basically though it comes down to price, and whether better efficiency and a slight edge in most current games is equally or more important than the possible better aging of GCN. So after hundreds of posts and partisan arguing in multiple threads, it is basically a toss up.

It's not often we agree, but there it is. It's actually a toss up right now and which one you get, depends on what price you can get them for really.

It's definitely not as bad for AMD as the 960 vs 380 scenario, where the massive power usage delta meant they are ruled out for low-power systems.

 

[3DVagabond]

Awesome Asus cooler,LOL!!

https://pics.computerbase.de/7/3/5/3/1/14-1080.2245261692.jpg

Edit to post!!

Supposedly,the overclock was on stock voltage.

You don't know what you'll start. lol There are some people who think not contacting the heatpipes doesn't matter.

 

[sirmo]

You don't know what you'll start. lol There are some people who think not contacting the heatpipes doesn't matter.

I don't think it matters that much personally.. they are still connected.. it could be better, but those pipes not touching the die are still dissipating heat.

 

[3DVagabond]

No, I'm serious. It wasn't meant in jest. I haven't compared pcb boards on either, but did notice an 8pin vs. 6pin. I'll be taking my reference 480 apart here soon, at least when usps delivers it.

Sorry, maybe I over reacted.

If you look at Powercolor's website they show it's different power delivery with stronger and more VRM's. Of course 8 pin, and the complete cooler is different not just the fans. All of this requires a different PCB too.

 

[3DVagabond]

One can argue all day about how the calculate the results, but bottom line is that the performance is pretty much a draw. I think one would have a hard time telling the difference in the gameplay in nearly any of the games if he were playing blinded to which card was in use. 1060 does have an edge in the 3 lowest performing games where the games are borderline playable, but more playable on the 1060.

Basically though it comes down to price, and whether better efficiency and a slight edge in most current games is equally or more important than the possible better aging of GCN. So after hundreds of posts and partisan arguing in multiple threads, it is basically a toss up.

I wouldn't call it a toss up. It's clearly what is more important. DX11 or DX12/Vulkan performance. Some will care about power usage. With DX12/Vulkan though you don't have to O/C your CPU to shift the bottleneck to the GPU. That saves power if people honestly care.

 

[Madpacket]

Yup 1060 - rx480 are about equal in performance today (AIB vs AIB). Certanly not a noticeable difference anyone would notice in most games.

rx480 pros:
- more VRAM
- FreeSync support (more choice and cheaper)
- crossfire support on the off chance you might want to get a second one at some point
- should perform better with most future titles
- slightly cheaper
- decent crypto miner if that's your thing

1060 pros:
- better legacy game performance, so if you play old games or a specific gameworks title a lot it might be a better choice
- 30 less watts of power used on average..

I like those pro's list:

Let me add a few:

rx480 pros:

-4GB card save you $40-50.00 over a 1060 which in most scenarios won't matter. 4GB really is enough for this card given the fill rate capabilities. Also current 4GB cards are really just 8GB cards and a simple reflash will remove this artificial limit if this really bothers you. This won't last forever but this makes it a great buy if you can get your hands on one now.

-Not just a decent crypto miner, the Rx40 currently the best crypto miner (hashes per watt) and probably is the main factor in why they're completely sold out at the moment (or sell out within minutes of availability).

1060 Pros:

-Form factor, you can already buy ITX sized (evga, zotac) 1060's which cost the same or less than the 8GB Rx480's. This is great for small form factor PC's.

-Power consumption (already mentioned) but can't be overstated for small form factor PC's. This is especially important as the ITX sized cards are not blower style so the heat will end up inside the case. When working with 12L or less sized cases this makes a huge difference on what type of CPU you can use or how far you can push the rest of your components.

 

[poofyhairguy]

Basically though it comes down to price, and whether better efficiency and a slight edge in most current games is equally or more important than the possible better aging of GCN. So after hundreds of posts and partisan arguing in multiple threads, it is basically a toss up.

Well said.

/thread

 

[biostud]

Anyone know anything about sub $399 amd videocards based on 14nm? Is it cut down version of vega? Suspecting vega will be in the +$599 category.

 

[3DVagabond]

Anyone know anything about sub $399 amd videocards based on 14nm? Is it cut down version of vega? Suspecting vega will be in the +$599 category.

There are supposed to be two Vega chips. Vega 10 and 11 with 10 being the smaller one this time. I believe Vega 11 will be in the $600 price range, but Vega 10 should be somewhere around that $400 price point.

 

[Zargon]

One can argue all day about how the calculate the results, but bottom line is that the performance is pretty much a draw. I think one would have a hard time telling the difference in the gameplay in nearly any of the games if he were playing blinded to which card was in use. 1060 does have an edge in the 3 lowest performing games where the games are borderline playable, but more playable on the 1060.

Basically though it comes down to price, and whether better efficiency and a slight edge in most current games is equally or more important than the possible better aging of GCN. So after hundreds of posts and partisan arguing in multiple threads, it is basically a toss up.

why do people keep using 1080p charts for comparos, this isnt 2010

they obviously kill it at 1080, wheres the 1440 tables

 

[zinfamous]

why do people keep using 1080p charts for comparos, this isnt 2010

they obviously kill it at 1080, wheres the 1440 tables

Because that is the vast bulk of the market, thus far more relevant to more consumers.

Besides, it's only really been about 2 generations that even high end cards have been able to dependably push 1080 at the highest settings in the best-looking games. This current generation seems to have only brought mid and in some cases--low end cards into that territory, so it is still very relevant.

Yes--1440p and 4k comparisons are great to look at what some of these cards can do, but it still isn't all that relevant to the market. Nothing can push 4k--probably the newest Titan, but who the hell is going to buy that? Less than 0.5% of the market, and that is probably being generous.

Only a handful of high-end cards can do wonders at 1440 right now, so it's still niche territory. Now, with DX12 optimizations and especially Vulkan, it seems possible to squeeze much more performance when these APIs are optimized for the games that devs choose to optimize them for.

 

[Zargon]

It will help show the cards longevity. If you expect to use the card for 2 years or so, and need a screen in that time, 1440p is clearly an option

but mostly, was joking

though really, half the games on the chart were over 100FPS

time to raise the bar....