AMD Radeon R9 480 3DMark11 benchmarks

AMD Radeon R9 480 3DMark11 benchmarks from VideoCardZ:

Please note: codenames aka ‘Driver Names’ were shown before Futuremark removed this line from result page. Results vary by 20% while 3DMark shows the same clock, so lower scores were either run with older, unsupported drivers, or higher scores simply show overclocked scores. You are the judge here. I didn’t want to post this because I simply could not confirm which results are showing stock performance.

In the following table I gathered the most plausible results. Chart shows the best possible scenario based on those results.

----

So the performance was expected to be around Fury nonX.

Now they need to price it agressively like $250 and it will sell a lot...

in cases like these where its speculation I always take the lowest/worst option/estimate and if proves to be better in the end then great.

so if the lowest score is that of a stock 480 on unsupported drivers, the fact that it beats the 970 is good news. now if as you theorize they can price it around $230 - $270 that would be excellent, bad news is, nvidia will look at it and respond with a similarly price performance 1060.

Why they using a totally out of date bench??

@urbangamez: Its not so much speculation, since the benchmarks are real and they have a legit link to 3DMark site.

Its more like if this is an engineer sample or a consumer ready GPU.

As I said its expected to be like Fury performance and not more, since we also know that this chip will be the flagship mobile chip.

Last time AMD did a similar thing, 7970M was a low clocked 7870 with near 7850 performance.

I think too an R9 480X would be nice at $280 ($100 lower than GTX1070) and R9 480 for $220.

ps: These prices are for the best possible scenario. If the worst scenario will come, I think that AMD will have problems selling any of these cards...

I have no idea and I have the same question.

3DMark should bury 3DMark11 in the same place with ET for Atari cartridges. And leave it there!!!

That's no good , they dug those up

This time we will find a better spot!!! :P

Ehhh well thats disappointing. Gtx 1070 for me than?

This is interesting.

That 480x in Crossfire

Graphics Score

25 803

Physics Score

9 014

Combined Score

8 142

My new Titan X (Link)

Graphics Score

28 670

Physics Score

14 839

Combined Score

14 119

Smashes it.

These cards better be cheap as they're not worth much already.

@klunt_bumskrint: Yeah they are not even near a nice o/ced Titan X when they are CFed.

Now that you mentioned it, I think with this performance $280 might be too much. Maybe, as I said in the begining, $250 is the perfect price...

This is likely why AMD was complaining about 3DMark being so outdated. At Capsaicin, they showcased Polaris 10 maintaining 60fps while running Hitman @1440p/60Hz using max detail settings. That would put it on par with, or atleast equal to, Titan X while running what was likely the only game that the drivers were ready for at the time.

That are some seriously low core clocks... And awful results.

There's a mobile Polaris 11 with 1395 Mhz clock speed.

Polaris 10 with 1395 Mhz is estimated to have 19900 score which is slightly faster than Fury X's 19315 score. Polaris 10's 40 CU at 1395 Mhz yields 7.142 TFLOPS which is similar to Fury Pro's TFLOPS.

A reminder from AMD's March 2016 road map

From https://developer.nvidia.com/dx12-dos-and-donts

Quote from NVIDIA...

"On DX11 the driver does farm off asynchronous tasks to driver worker threads where possible"

NVIDIA DX11 drivers are already using key DX12 style speed up methods i.e. asynchronous tasks and multithreading.

Under DX12, AMD gains these speed-up methods.

@ronvalencia said:

There's a mobile Polaris 11 with 1395 Mhz clock speed.

Polaris 10 with 1395 Mhz is estimated to have 19900 score which is slightly faster than Fury X's 19315 score. Polaris 10's 40 CU at 1395 Mhz yields 7.142 TFLOPS which is similar to Fury Pro's TFLOPS.

This image is from previous oldest article.

It has nothing to do with the new article that was published yesterday...

@Coseniath said:

@ronvalencia said:

There's a mobile Polaris 11 with 1395 Mhz clock speed.

Polaris 10 with 1395 Mhz is estimated to have 19900 score which is slightly faster than Fury X's 19315 score. Polaris 10's 40 CU at 1395 Mhz yields 7.142 TFLOPS which is similar to Fury Pro's TFLOPS.

This image is from previous oldest article.

It has nothing to do with the new article that was published yesterday...

The image refers to mobile parts.

Polaris 10's clock speed started from 800 Mhz to 1050 Mhz to 1266 Mhz. Again, refer to AMD's March 2016 road map.

Polaris 10 with 36 CU at 1266 Mhz (5.85 TFLOPS) has effectively replaced R9-390 (5.9 TFLOPS) as per AMD's March 2016 road map.

One more thing, according to Sisoft database, C7 has 2304 cores, there is a possibility that there's another P10 chip with 2560 cores (full silicon), even faster than those. Just wanted to make it clear.

Just throwing it out there. Leaning heavily towards 480/480x in speculation now and no high end at all, but still theres hope that the p10 could have a gtx1070 fighter.

Polaris 10 36 CU 1266 Mhz scored 18060.... 40 CU at 1266 Mhz version could score 20,066 points.

Polaris 10 40 CU at 1395 Mhz estimated score could be 22,111 points.

Well, we don't disagree here.

And I think Polaris 10 36 CU will perform much better than R9 390.

In gaming especially when we compare different architectures TFLOPs matters not...

3DMarks is useful for comparing within similar GPU design and drivers e.g. R9-390X is lesser GCN than R9-Fury X.

Atm, Polaris 10 36 CU(C7) at 1266Mhz configuration doesn't full fill AMD's March 2016 road map's replacing Fury X requirement.

Again,

WhyCry VideoCardz.com • a day ago

One more thing, according to Sisoft database, C7 has 2304 cores, there is a possibility that there's another P10 chip with 2560 cores (full silicon), even faster than those. Just wanted to make it clear.

@ronvalencia: Noone said that full Polaris10 is not a 2560 cores chip.

But that doesn't mean it will be this generation's AMD high end.

So you are saying that a next gen architecture flagship chip with double fabrication jump will be in the same performance as previous gen?

Do you think that AMD has plans to point 232mm2 chip as flagship replacer of FuryX?

Having the same performance as a former architecture flagship doesn't mean replacement for a new chip.

Was 7870 a replacement to 6970 since it has the same and a little more performance? Of course not. 7970 was.

Vega will be high end GPU from AMD.

"Polaris will be a mainstream GPU."

-Thus spoken AMD's Roy Taylor.

"Mainstream GPU" refers to the price for the GPU.

Polaris 10's performance range covers current gen AMD GPUs which ranges from R9-390X to Fury X as per AMD's March 2016 road map.

Polaris 10's 232 mm^2 at 14 nm is effectively 464 mm^2 at 28 nm. To match Fury X's 590 mm^2, higher clock speed makes up the the missing CUs in Polaris 10.

To go beyond Fury X's performance level, that's Vega 10.

Polaris 10 36 CU at 1266 Mhz yields 5.83 TFLOPS

R9-390X 44 CU at 1050 Mhz yields 5.93 TFLOPS

-------

Polaris 10 40 CU at 1350 Mhz yields 6.91 TFLOPS

R9-Fury Pro 56 CU at 1000 Mhz yields 7.1 TFLOPS

-------

Polaris 10 40 CU at 1570 Mhz yields 8.04 TFLOPS

R9-Fury X 64 CU at 1050 Mhz yields 8.6 TFLOPS

---------

Vega 10 (2X from mobile Polaris 10's 32 CU 1350Mhz) = 11 TFLOPS.

@ronvalencia: We are saying the same thing with different words. :P