ASUS Radeon R9 380X Strix 4GB

[Steevo]

What? I'm using one at 2560 at reduced settings. Ultra is for benchmark articles. Of course the 960 does 1080 fine, now you're just out there. Not replying to any more troll replies.

So a 960 owner who comes to threadcrap about a card that whips their cards ass, then clearly states they are too lazy to provide proof to backup their claims, and when is provided said proof goes off on a tangent isn't trolling? Trolling would be me telling you my 7970 kicks your card square in the balls for any reason other than it really does, and it cost me less.


The point of benchmarking is to see where the strengths and weaknesses are, not to prop up a flacid dick so anyone can feel better and get some confirmation bias feels by adjusting games to make sure it always looks the best. By having all the games at their highest settings W1zz promotes a few core values, first, the unwillingness to compromise on quality, the second to reinforce a even and level playing field, third to help push the envelope on hardware/software fronts that benefit all of us.

A shit attitude of "well its runs good on medium or low settings" is equal to "Well it looks as good as a potato box 4" and defeats the purpose of an enthusiast site, where we set high benchmark today to become the standard for tomorrow.

 

[MagnuTron]

I have the R9 380 (without x) version of this card and everything looks identical, box, pcb, everything. Its a good card, but please check the VRM temps. On my card it goes to >120 degrees Celsius in Furmark, while on Heaven loop it sits on 100 Celsius. While gaming occasionally reaches 100, but in average it's a bit under. All this on stock, not even the OC profile which comes OOB.

I contacted Asus and they told me they have high quality components ... yala yala yala, and its perfectly fine for this model ... go figure. Somehow, as there was actually nothing wrong with it and it was working perfectly, I did not return it.

Anyway for this reason, I'd definitely not give it the "Highly Recommended" mark, at least not to this Asus. R9 380X is great and I saw on other brands, like Saphire a solution which had VRMs touching the big heatsink so for sure they didn't had this problem.

I just built a 380 machine for a customer yesterday. Sadly the cards dont have dual bios. Otherwise a flash upgrade would of been a cool thing to try out huh?

EDIT: Typo0z

 

[W1zzard]

TPU's 960 numbers on this test don't agree with TPU's 960 numbers on prior tests.

Two games have huge performance issues on Windows 10 on 2 GB NVIDIA cards.
GTA V and Watch Dogs.
This seems to be an NVIDIA driver issue, on Windows 7 everything was fine, using the same settings.
AMD 2 GB cards are unaffected too.

 

[Cataclysm_ZA]

It has been noted around the web that TPU's chart seems to suddenly show the 960 a bit lower, tied with 270X. That's what rruff is getting at. I'm too lazy to verify this though.

Yes I did. Your post is irrelevant though. TPU's 960 numbers on this test don't agree with TPU's 960 numbers on prior tests. It's a typo.

Have you guys ever considered that because W1zzard re-does the benchmarks for all the supported GPUs when new drivers come out, that the numbers will change over time to reflect how much difference the drivers can make? Here's an example of the GTX 960's position relative to the Zotac GTX 980 Ti AMP:

Here W1zzard's using a Skylake build, Geforce WHQL 358.50, Catalyst 15.9.1 beta, Windows 10. The grouping is similar for the MSI GTX 980 Ti Lightning review, though the results taken with the Zotac review show a 1% performance gain for both cards. This could be through various games getting patches to improve performance, for example, or because Windows 10 in the Zotac review could be build 10586, which includes the Speed Shift optimisations for Skylake processors.

Now, compare that to the results taken when the last GTX 950 was reviewed:

This time, he's using a Haswell build, Windows 7 SP1, Geforce WHQL 353.06, 355.65, and CCC 15.5 beta

By his own admission, under the listing for the test system, he says: "Benchmark scores in other reviews are only comparable when this exact same configuration is used." That's why the results suddenly show the GTX 960 tying with the R7 270X, because not only are the Windows 10 drivers showing a performance boost, there's also a faster CPU in there, different memory speeds, and a different OS.

 

[newtekie1]

Yes I did. Your post is irrelevant though. TPU's 960 numbers on this test don't agree with TPU's 960 numbers on prior tests. It's a typo.

No they aren't. The numbers have changed due to rebenching with a different test setup and different games being added and removed from the test suite. And the new drivers introduced quite a few driver bugs for the 960. So it's overall performance went down. This happens pretty often, AMD had issues similar to this before too.

 

[mab1376]

This was not the purpose of this card, even when they launched 285, the goal was to get pretty much the same performance as previous 280 generation, that's why the name was in the middle of those 2, while having lower production cost. And this card does exactly that, gets the same performance with 256 bus, instead of 384 bit, which is a good achievement from AMD. Also brings more recent GCN, with HW decode for 4k, Freesync, True Audio and other things.

Just seems confusing to consumers. Seems like it would be the next generation; not a 280xv2

 

[Cataclysm_ZA]

Just seems confusing to consumers. Seems like it would be the next generation; not a 280xv2

It's a bit like the HD5870-HD6870 switch. A slight regression in performance with better power savings, yet the model number implied that it was faster.

 

[Tsukiyomi91]

So a 960 owner who comes to threadcrap about a card that whips their cards ass, then clearly states they are too lazy to provide proof to backup their claims, and when is provided said proof goes off on a tangent isn't trolling? Trolling would be me telling you my 7970 kicks your card square in the balls for any reason other than it really does, and it cost me less.


The point of benchmarking is to see where the strengths and weaknesses are, not to prop up a flacid dick so anyone can feel better and get some confirmation bias feels by adjusting games to make sure it always looks the best. By having all the games at their highest settings W1zz promotes a few core values, first, the unwillingness to compromise on quality, the second to reinforce a even and level playing field, third to help push the envelope on hardware/software fronts that benefit all of us.

A shit attitude of "well its runs good on medium or low settings" is equal to "Well it looks as good as a potato box 4" and defeats the purpose of an enthusiast site, where we set high benchmark today to become the standard for tomorrow.

True. Also, no one in their right mind would say that a sub $200 card has the ability to cook a much more expensive competitor (e.g GTX970 G1 Gaming) that sports a beefier bus width, faster chip & consumes a lot more power with increased power limit while built on a custom PCB + cooler to let the chip go further than the reference model for a few extra bucks. Benches done by W1zz is considered as even, no holds barred & BS-free. Results are absolute. Just take in some salt & accept the truth.

 

[EarthDog]

Are you suggesting that he retests all the old cards in the list when there is a config change?

I believe he does that...

 

[Cataclysm_ZA]

Are you suggesting that he retests all the old cards in the list when there is a config change? Or that W10 drivers suddenly crippled the GTX 960 relative to everything else?

He's done that for just over a decade, if I'm not mistaken. He once discussed the amount of work that goes into a GPU review on these forums a long time ago. I remember the math worked out to being tied up for three weeks straight of just testing old cards if he didn't have multiple rigs running at the same time to speed things up.

Edit: And yes, Windows 10 drivers are doing something wonky to the 2GB Kepler and Maxwell cards, specifically when it comes to GTA V.

 

[rruff]

Two games have huge performance issues on Windows 10 on 2 GB NVIDIA cards.
GTA V and Watch Dogs.
This seems to be an NVIDIA driver issue, on Windows 7 everything was fine, using the same settings.
AMD 2 GB cards are unaffected too.

Yikes! Do you know any more detail on what exactly is happening? Nvidia really screwed that up...

 

[rruff]

I believe he does that...

Yes, I read his post, so deleted mine.

 

[W1zzard]

He's done that for just over a decade, if I'm not mistaken. He once discussed the amount of work that goes into a GPU review on these forums a long time ago. I remember the math worked out to being tied up for three weeks straight of just testing old cards if he didn't have multiple rigs running at the same time to speed things up.

That is correct, I rebench all cards for every iteration of the test system (as indicated by the version number in the test setup header, currently at 40). It's roughly two weeks now for a whole rebench. Using just a single rig, so there are no random deviations

Yikes! Do you know any more detail on what exactly is happening? Nvidia really screwed that up...

I have no idea, my best guess is it has to do with Windows 10 GPU memory management. But since AMD is not affected, I can't be sure of that

 

[rruff]

I have no idea, my best guess is it has to do with Windows 10 GPU memory management. But since AMD is not affected, I can't be sure of that

Not seeing any mentions of this online, but it sure is worthy of mentioning. Did you try different drivers?

And holy crap, I didn't realize you retested every card! Thought you would just test a couple and scale the rest, for the summary anyway.

Just had another look at the chart you posted above for GTA 5 and the R9 285 is also effected, though not as much as the Nvidia cards.

 

[xorbe]

Okay, so I'm glad I didn't move to Win10, heh. At least now we know the rest of the story, and why other reviews had different numbers than TPU.

Steevo, click System Specs over there (<-) before flying off the rails next time, pretty please. Was just casually commenting on the 960 comments, as I'm testing one right now in my godbox system, for one of my HTPC setups.

W1zzard, big thanks for the reviews of course. I know it takes a lot of time and work, and I read it for free, so I have no room to complain.

 

[Casecutter]

Send me to the sites your on, I cannot find anything close to those prices even with rebates

Well I suppose there's some one off flash, or screwed-up listed price that got by on Jet.com, but that's hardly the norm. I use the charts from pcpartpicker.com to kind of determine the trends. But correctly there quite a few 380 2Gb and sell-off of 960 2Gb that have been as low as $150 for a PowerColor 380 2Gb working a $20 rebate, while 960's 2Gb have gotten to like $160 -AR leaving 950's high and dry.

Also, the 960 is a budget card that replaces the 750Ti due to it's shortcomings (lack of 6-pin connector, capped power limits etc),

Sorry don't mean to bust your... but the 960's can out as a 760 replacement and a card to capture the older GTX660 crowd or those looking to the R9285 The 750Ti those weren't "shortcomings" those were touted as features when it came out and still are, such a card is in a completely lower rung "entry-gaming". The 960 was a mid-range for the "sweet spot" for running MOBA games something GTX 750Ti didn't provide in spades.

http://www.geforce.com/whats-new/articles/introducing-the-geforce-gtx-960

As to this 380X well not the boost one would consider from a XT variant but this is the first review. There was more than 10% from a 280-280X if I recall. So this kind of spells out why AMD held out this long, not good to release a new card when 280X/290 was still abundant in the market space.

That said, would still look at one for someone's new build if I was looking in the next couple months especially if the toss a $20 rebate and a game or two in the mix.

 

[N3M3515]

This card is very good for the price.
The thing that bugs me is . . . . . .
Why the hell are the gtx 770 and 780 so slow???
The gtx 770 used to be neck and neck with the 7970 GHZ Ed. if not edging it out (and the ghz ed is faster than the 280x!). Now it gets beated by basically a 7950???
And the gtx 780 Ti was always faster than the 290x. Specially at lower resolutions.
Someone please an explanation
Humansmoke where are you

 

[rtwjunkie]

This card is very good for the price.
The thing that bugs me is . . . . . .
Why the hell are the gtx 770 and 780 so slow???
The gtx 770 used to be neck and neck with the 7970 GHZ Ed. if not edging it out (and the ghz ed is faster than the 280x!). Now it gets beated by basically a 7950???
And the gtx 780 Ti was always faster than the 290x. Specially at lower resolutions.
Someone please an explanation

I can help you on that. Nvidia stopped optimizing Kepler with 347.88. Meanwhile, driver refinement has continued with Maxwells. We shall see the same thing happen in about 8 to 10 months after Pascal starts supplanting Maxwell sales.

 

[xorbe]

This card is very good for the price.
The thing that bugs me is . . . . . .
Why the hell are the gtx 770 and 780 so slow???
The gtx 770 used to be neck and neck with the 7970 GHZ Ed. if not edging it out (and the ghz ed is faster than the 280x!). Now it gets beated by basically a 7950???

Driver has Win10 problem for 770/960 at least, that's hidden in the summary chart. That's what the hubbub above was about.

 

[newtekie1]

I have no idea, my best guess is it has to do with Windows 10 GPU memory management. But since AMD is not affected, I can't be sure of that

It is very interesting that it just caps the framerate for 2GB cards right at 12FPS, regardless of the card. Hopefully nVidia gets a handle on this in a future drive update.

 

[N3M3515]

I can help you on that. Nvidia stopped optimizing Kepler with 347.88. Meanwhile, driver refinement has continued with Maxwells. We shall see the same thing happen in about 8 to 10 months after Pascal starts supplanting Maxwell sales.

So gtx 770/780 users are fucked....for new games
I was so close to buy a gtx 780 used but the difference is nothing...

 

[rtwjunkie]

So gtx 770/780 users are fucked....for new games
I was so close to buy a gtx 780 used but the difference is nothing...

In reality, I was using a 780 until 2 months ago. It did not suffer any in real life gaming use. Benchmarks like W1zzard's tests are set an an abnormally high standard for uniformity. The difference is mostly only visible in extreme benchmarks. You would be fine.

 

[xorbe]

Similarly, 380X + Win10 seems to stumble in Metro Last Light. The core clock drops down. I guess this is similar to when nvidia users randomly need to choose "prefer maximum performance" in the control panel for basically the same issue.

http://www.legitreviews.com/sapphire-radeon-r9-380x-nitro-video-card-review_175481/7

 

[Fluffmeister]

True, I guess I kinda overhyped this card

No doubt, Tonga is huge compared to the GM206 powered GTX 960, gotta love those margins.

 

[EarthDog]

Similarly, 380X + Win10 seems to stumble in Metro Last Light. The core clock drops down. I guess this is similar to when nvidia users randomly need to choose "prefer maximum performance" in the control panel for basically the same issue.

http://www.legitreviews.com/sapphire-radeon-r9-380x-nitro-video-card-review_175481/7

weird.. can't say I noticed that in my review (not published).. let me see if I can capture some data and post it...

EDIT: Disregard.. I somehow missed W10.. we stillare on W7..