Windows 10, Nvidia being left in the dust.

TaintedSquirrel

TaintedSquirrel

[H]F Junkie
Joined
Aug 5, 2013
Messages
12,388
TechPowerup has updated their test suite to Windows 10 this week, so I took the time to compare results. These are canned benchmarks.

Here's their previous GPU review from September, still running Windows 7:
http://www.techpowerup.com/reviews/Gigabyte/GTX_950_Xtreme_Gaming/
1080p averages: http://i.imgur.com/n95kC6O.gif

This is their most recent review, the first with Windows 10 & Skylake build, they also added Mad Max and MGSV to their test bench:
http://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Lightning/
1080p averages: http://i.imgur.com/mqXxy9v.png

The newest review also includes a driver update to 15.9.1.

370 gained 18%, putting it on-par with the GTX 760 & 950.
270X gained 17%, putting it on-par with the GTX 960.
285 gained 8%, putting it slightly ahead of the 770.
280X gained 18% over the 770.
290 gained 5%, putting it on-par with the GTX 970.
Fury X gained 7%, putting it about 5% away from the 980 Ti.
295X2 gained 6% over the Titan X.

So on average, GCN 1.0 chips gained nearly 20%, GCN 1.1 and 1.2 around 5-10%.
 
Last edited:
Good for AMD! Looks like there is some actual competition, now. I will say that the Windows 10 drivers from NVIDIA are still questionable, though. I would love to see a Windows 10 Performance review from [H].
 
Windows 10 Drivers from AMD are indeed very good! I really like how my 280x cards performance keeps increasing, a 20% increase is just plain AWESOME! Bloody cards went up at least 1 performance tier since I purchased them.

I rather like rebadging cards personally, since it comes with longer active support! :)
 
So I guess the take-away here is that when you buy nvidia you get the performance up front. And when you buy AMD you need to wait five years for drivers to catch up.

Am I reading this right?
 
jwcalla said:
So I guess the take-away here is that when you buy nvidia you get the performance up front. And when you buy AMD you need to wait five years for drivers to catch up.

Am I reading this right?
Click to expand...

Wait for the OS and game developers to catch up. :)
 
jwcalla said:
So I guess the take-away here is that when you buy nvidia you get the performance up front. And when you buy AMD you need to wait five years for drivers to catch up.

Am I reading this right?
Click to expand...


Please refer to one card on that list that came out five years ago. All of them came out within the past year or less.

AMD probably developed these cards with dx12 in mind.

Please go.
 
TorryHolt81 said:
Please refer to one card on that list that came out five years ago. All of them came out within the past year or less.

AMD probably developed these cards with dx12 in mind.

Please go.
Click to expand...

all of the GCN 1.0 cards are 4 years old. with the exception of 290 and 290X all of those R9 200 and R7 200 are rebranded HD7000 cards.
 
TorryHolt81 said:
Please refer to one card on that list that came out five years ago. All of them came out within the past year or less.

AMD probably developed these cards with dx12 in mind.

Please go.
Click to expand...

Pitcairn is nearly 5 years old at this point, and Kepler is almost 4 years old. Both are represented in the list.
 
TaintedSquirrel said:
TechPowerup has updated their test suite to Windows 10 this week, so I took the time to compare results. These are canned benchmarks.

Here's their previous GPU review from September, still running Windows 7:
http://www.techpowerup.com/reviews/Gigabyte/GTX_950_Xtreme_Gaming/
1080p averages: http://i.imgur.com/n95kC6O.gif

This is their most recent review, the first with Windows 10 & Skylake build, they also added Mad Max and MGSV to their test bench:
http://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Lightning/
1080p averages: http://i.imgur.com/mqXxy9v.png

The newest review also includes a driver update to 15.9.1.

370 gained 18%, putting it on-par with the GTX 760 & 950.
270X gained 17%, putting it on-par with the GTX 960.
285 gained 8%, putting it slightly ahead of the 770.
280X gained 18% over the 770.
290 gained 5%, putting it on-par with the GTX 970.
Fury X gained 7%, putting it about 5% away from the 980 Ti.
295X2 gained 6% over the Titan X.

So on average, GCN 1.0 chips gained nearly 20%, GCN 1.1 and 1.2 around 5-10%.
Click to expand...

Dude these are normalized percentages, you can't take the change in percentage, as equal because the normalization factor is different ;), I'm surprised no one else caught that so far.

If you need to look at the data before the normalization is there.
 
Araxie said:
all of the GCN 1.0 cards are 4 years old. with the exception of 290 and 290X all of those R9 200 and R7 200 are rebranded HD7000 cards.
Click to expand...

Interesting... Shows how much I know about AMD.

I withdraw my previous statement.
 
jwcalla said:
So I guess the take-away here is that when you buy nvidia you get the performance up front. And when you buy AMD you need to wait five years for drivers to catch up.

Am I reading this right?
Click to expand...

Actually they were price competitive when they came out, 4-5 years later their performance actually increased making them a better RoI than their green counterparts.
 
TaintedSquirrel said:
370 gained 18%, putting it on-par with the GTX 760 & 950.
270X gained 17%, putting it on-par with the GTX 960.
285 gained 8%, putting it slightly ahead of the 770.
280X gained 18% over the 770.
290 gained 5%, putting it on-par with the GTX 970.
Fury X gained 7%, putting it about 5% away from the 980 Ti.
295X2 gained 6% over the Titan X.

So on average, GCN 1.0 chips gained nearly 20%, GCN 1.1 and 1.2 around 5-10%.
Click to expand...

I chose, at random, to look at the Crysis 3 page in the Win7 review and the Crysis 3 page in the Win10 review, and looked at the 1080p scores:

370 Win7: 22.2 FPS. 370 Win10: test not performed.
270X Win7: 25.5 FPS. 270X Win10: test not performed.
285 Win7: 28.7 FPS. 285 Win10: test not performed.
280X Win7: 32.9 FPS. 280X Win10: test not performed.
290 Win7: 42.9 FPS. 290 Win10: test not performed.
Fury X Win7: test not performed. Fury X Win10: 57.3 FPS.
295X2 WIn7: test not performed. 295X2 Win10: 63.1 FPS.

Okay... looking around at the other game tests, I realize there are no AMD video cards from your list that were tested in both the Win7 and Win10 reviews.

Whatever it is that is being compared to get those performance increases, it is not the reviews you linked to, so what tests are being compared?
 
all i know is that nvidias drivers kinda suck on win10. ive been stuck with a bug since launch day. if i launch and then minimize or exit any dx10/11 game, the gama goes up like 30%. it has to be reset before i maximize the game again.
 
TaintedSquirrel said:
TechPowerup has updated their test suite to Windows 10 this week, so I took the time to compare results. These are canned benchmarks.

Here's their previous GPU review from September, still running Windows 7:
http://www.techpowerup.com/reviews/Gigabyte/GTX_950_Xtreme_Gaming/
1080p averages: http://i.imgur.com/n95kC6O.gif

This is their most recent review, the first with Windows 10 & Skylake build, they also added Mad Max and MGSV to their test bench:
http://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Lightning/
1080p averages: http://i.imgur.com/mqXxy9v.png

The newest review also includes a driver update to 15.9.1.

370 gained 18%, putting it on-par with the GTX 760 & 950.
270X gained [b]17%[/b], putting it on-par with the GTX 960.
285 gained 8%, putting it slightly ahead of the 770.
280X gained 18% over the 770.
290 gained 5%, putting it on-par with the GTX 970.
Fury X gained 7%, putting it about 5% away from the 980 Ti.
295X2 gained 6% over the Titan X.

So on average, GCN 1.0 chips gained nearly 20%, GCN 1.1 and 1.2 around 5-10%.
Click to expand...

Math fail?
 
pxc said:
Yeah, I also found the OP's analysis "iffy"*, as the posters on reddit did earlier. lol https://www.reddit.com/r/Amd/comments/3qvth3/techpowerup_finally_upgraded_to_windows_10/

* by iffy, I mean completely nonsensical
Click to expand...
TechPowerup is blacklisted there.
Along with KitGuru, TechReport, AnandTech, HardOCP, and PCPer.

Whatever it is that is being compared to get those performance increases, it is not the reviews you linked to, so what tests are being compared?
Click to expand...
Contact wiz and ask him when he performed the benches. It's not my responsibility to validate his numbers.
 
pxc said:
That's not the problem with the claims you're making. Nice deflection attempt though. lol
Click to expand...
Except their claim is that TechPowerup is Nvidia-biased, despite the numbers favoring AMD. You can't claim TPU is biased for Nvidia when they're showing up to 20% leads on AMD hardware. Can't even believe I have to say this right now.

As for the benchmarks' accuracy, I pulled their summaries and calculated the results. I can't say anything else about their testing methods. They could be wrong.
 
TaintedSquirrel said:
Contact wiz and ask him when he performed the benches. It's not my responsibility to validate his numbers.
Click to expand...

TaintedSquirrel said:
Very true.
So when AMD does something right, it's even harder to notice.
Click to expand...


There is nothing done right there, you are not reading the % right because the normalization skews the numbers based on a factor that isn't equal to both test suites. If you can't understand that, just go through the raw data and see for your self, there is very little change in % differences for AMD and nV cards.
 
razor1 said:
normalization skews the numbers
Click to expand...
Normalization doesn't "skew" anything, it's just a scale. I unnormalized the data and calculated the % change on each AMD card relative to its Nvidia counterpart (listed in the post). If you'd like a basic math explanation on how that works I would be happy to provide it. It would take about 30 seconds for anyone with a calculator to verify my numbers, should be an error of about 1% since I hastily rounded.

The numbers are accurate, TPU's benchmarks may not be.
 
TaintedSquirrel said:
Normalization doesn't "skew" anything, it's just a scale. I unnormalized the data and calculated the % change on each AMD card relative to its Nvidia counterpart (listed in the post). If you'd like a basic math explanation on how that works I would be happy to provide it.
Click to expand...

Good. Do it.
 
I am in the market for the 970itx or the Fury Nano... But looking at the Nano OC vs the 980 (970 OC equivalent) it's around 100% more cost for 10% performance improvement. I am not one to care for value curves but that's a tough one to swallow.

Why do all these review leave off the 970? It's the main competitor... Give a 970 itx vs Fury Nano OC vs OC review and it'd be a slaughter.

I do like the Fury X but I don't think there's a way to fit it into an Osmi. I won't have that case to tell until December.

I think both the lineups are pretty competitive. You generally won't notice 10%. The only outlier is the Nano it seems...
 
Last edited:
Brent, I think people are seeing the performance boost because they are going from 7 to 10. I have seen no boost because I got back into of gaming after years with my Xbox 360 when windows 8 was launched. No performance increase noticed here.
 
It's also possible the newer generation CPU is making a difference. DDR3 vs DDR4 bandwidth differences could account for a chunk of that if the drivers are frequently shuffling around a lot of data. That Skylake is also clocked 7% higher plus any architecture improvements. If drivers are holding you back that will probably help.

I'm not sure I've seen someone actually benchmark the effect of CPU memory bandwidth on GPU performance.
 
Anarchist4000 said:
I'm not sure I've seen someone actually benchmark the effect of CPU memory bandwidth on GPU performance.
Click to expand...

Actually latest Digital Foundry's articles about pc performance have shown that, with small but perceptible gains on DDR4 speed. It is weird that they picked it up vs more "technical" sites tbh, it surprised me.

Look here on their Skykake I3 6100 review, at least on the entry level memory performance matters with a notorious increase.
http://www.eurogamer.net/articles/digitalfoundry-2015-intel-core-i3-6100-review
 
Tainted I like you but you can't just plaster percentages like this. Also the test suite is different so really the averages are not comparable from a purely Maths stand point.
 
TaintedSquirrel said:
Except their claim is that TechPowerup is Nvidia-biased, despite the numbers favoring AMD. You can't claim TPU is biased for Nvidia when they're showing up to 20% leads on AMD hardware. Can't even believe I have to say this right now.

As for the benchmarks' accuracy, I pulled their summaries and calculated the results. I can't say anything else about their testing methods. They could be wrong.
Click to expand...

Off course... No, they aren't. /sarcarms

Untitled.jpg
 
Last edited:
wantapple said:
Off course... No, they aren't. /sarcarms

http://s15.postimg.org/j1a2hljmj/Untitled.jpg[/IMG]
Click to expand...

That's hilarious. But they were kind of telling it how it was.

"At this price point, the GTX 980 beats AMD's Radeon R9 290X in everything: price/performance, absolute performance, performance per watt, power consumption, heat, and noise"

You have to remember the 290x was really expensive until the 970/980 launched. Then became competitive again when it dropped to ~280. Now AMDs lineup is a little high (in the neglible range to me) but not as absurd as it was. I think I'll wait for the custom Nanos to make my 970 vs nano decision.
 
I do not believe that manufacturers are allowed to make custom R9 Nanos.

Asus just released a "special version" of the card - it's just a repaint.

48146_07_asus-teases-amd-radeon-r9-nano-white-custom-pcb.jpg
 
Vittra said:
I do not believe that manufacturers are allowed to make custom R9 Nanos.

Asus just released a "special version" of the card - it's just a repaint.
Click to expand...

Does that white paint make it even more expensive?
 
Dayaks said:
That's hilarious. But they were kind of telling it how it was.

"At this price point, the GTX 980 beats AMD's Radeon R9 290X in everything: price/performance, absolute performance, performance per watt, power consumption, heat, and noise"

You have to remember the 290x was really expensive until the 970/980 launched. Then became competitive again when it dropped to ~280. Now AMDs lineup is a little high (in the neglible range to me) but not as absurd as it was. I think I'll wait for the custom Nanos to make my 970 vs nano decision.
Click to expand...

witch makes no sense, since 290x wasnt meant to be against 980... and as right know 290x is better than 980 LOL
 
wantapple said:
witch makes no sense, since 290x wasnt meant to be against 980... and as right know 290x is better than 980 LOL
Click to expand...

That article you linked is from the 980 launch. At the time it all was true. Overall the 980 was 10% faster stock and 24% faster OC vs OC (based on the [H] review) at launch. They were in the same price bracket at the time of that article.

It was nVidia's top card and AMDs. Comparing them was natural.
 
You must log in or register to reply here.
Back
Top