AMD Radeon R9 380X Nitro Launch Review

[RedJaron]

I am dissapoint, I was expecting the full 384 mem bus and substantially higher performance. I had been hyping this for a while as the reason for NV to release the 960 ti's and it most certainly won't

Tonga has never had a 384-bit bus. 256-bit IS full for Tonga.

 

[turkey3_scratch]

There still does not seem to be any good card to purchase in the $220-300 range. 380X just not worth it. 960 and 380 are below $200 now. GTX 970 and 390 are above $300.

 

[RedJaron]

There still does not seem to be any good card to purchase in the $220-300 range. 380X just not worth it. 960 and 380 are below $200 now. GTX 970 and 390 are above $200.

I was hoping for more out of the 380X as well. It still leaves a huge performance and money gap. NVidia can really capitalize with a 960 Ti if they do it right.

 

[blazorthon]

So we're at least semi agreeing that, limiting the discussion to modern titles and looking forward, cards like the title articles 380x are most definitely NOT 1440p cards, and at best are 1080p cards that are already staring that limit in the face as well.

With the 680 being talked about in the past tense so soon, I wonder if I didn't even underestimate how long my 970 will last.

Anybody here think that dx12 and other such software improvements will help keep these cards relevant? Or are we still looking at a ~3 year upgrade cycle to just run games at decent settings, even with mostly top-of-the-line hardware today?

I agree that the 380X and similarly performing cards are only mid-ranged 1080p cards for modern games.

As always, that depends on how long you can tolerate lowering settings as more and more intensive games come out.

DX12 will help with the CPU stagnation more than anything. I don't think it will help cards stay relevant longer because the next generation of cards will also benefit from DX12. It may help push back the limit you mentioned, but that's difficult to know for sure in advance because the DX12 games could just as easily also get even more intensive features to mitigate the performance gains of DX12 over DX11.

The biggest factor might be whether or not the difficulty with smaller fab processes will continue and if it does, can AMD and Nvidia really make any more huge architectural improvements? Maxwell (and Nano) pushed hard on this with more aggressive and robust power control, but that has greatly diminishing returns as you try to take it further. Maybe stacked memory can be taken further with stacked GPUs? Food for thought.

 

[LookItsRain]

Was really looking foward to this card, but it fails to impress at this price.

 

[TopLuca]

Disappointing release to say the least , the most interesting part is how the 970 beats the 390 in almost in every game in this review as the 390 nitro in its own review beats the 970 in almost every title! Then why on Earth use a PowerColor PCS 390. Anyways that was off topic , bottom line : That release isn't really anything special.

 

[turkey3_scratch]

That's what I asked, I don't see why they used a Powercolor 390 and an MSI GTX 970. I don't know, but it just seems a bit unfair and possibly biased toward Nvidia. There is clearly an MSI 390 with the Twin FrozrV coolers just like the 970 that both should have been used for a direct comparison.

I don't ever see Nvidia lowering the 970 price more (well, it really has not lowered much in the first place, if not at all). The GTX 970 sells like hotcakes. It seems to be the go-to card from Nvidia. Nvidia might actually have been smart in leaving nothing in the $200-$300 price gap, and it was probably part of their strategy. That way, instead of taking the weaker 960, people will just spend the $330 for a 970.

I do not see a 960Ti being released, because it'll only make Nvidia lose money by making people purchase less 970s.

 

[RedJaron]

The cards used in the review are dependent upon which cards the reviewer has. You may not have noticed that AMD isn't putting out many reference boards lately, so reviewers are dependent on MSI, Sapphire, PowerColor, etc submitting models for review. They can only compare data they have tested. If they don't have a particular card, it's not always feasible to go out and buy one. This was also done by Igor, not Kevin, so the card each has is not the same as the other.

 

[eyehategod666]

Are these benchmarks really accurate - the Nano doesn't seem to have that much of an edge over the GTX 970 for being a $650 card, that's really confusing to me.

 

[RedJaron]

The Nano is a Fiji card that's been power restricted to operate better in ITX cases. It has the better architecture, but the clock is slower than you'll find in the full desktop Fury versions.

 

[TallestJon96]

Seems like a good card. Its nice that it fills out the gap between the 960/380 and 979/390. If you look At the BF4 graph, its clear that there is a big difference between these two tiers. If NVIDIA released a 960 ti with 1280 cudas and 4gb ram, we'd have the gap filled out nicely.

Its a great example of how the graphics card market changes over time. 4 years ago a powerful card was released for $500, and now you can buy a very similar card for $240, its slightly more powerful, has more VRAM, and is more efficient.

 

[TbsToy]

Seems like a good card. Its nice that it fills out the gap between the 960/380 and 979/390. If you look At the BF4 graph, its clear that there is a big difference between these two tiers. If NVIDIA released a 960 ti with 1280 cudas and 4gb ram, we'd have the gap filled out nicely.

Its a great example of how the graphics card market changes over time. 4 years ago a powerful card was released for $500, and now you can buy a very similar card for $240, its slightly more powerful, has more VRAM, and is more efficient.

How idiotic. Why does she NEED to have her gap filled anyway?
Sincerely,
Walt Prill

 

[spentshells]

I am dissapoint, I was expecting the full 384 mem bus and substantially higher performance. I had been hyping this for a while as the reason for NV to release the 960 ti's and it most certainly won't

Tonga has never had a 384-bit bus. 256-bit IS full for Tonga.

I've read it quite a few places.

http://wccftech.com/amd-tonga-gpu-die-shot-features-384bit-memory-controller-fiji-die-size-approximated-560mm2/

I don't doubt you but I read it a few different places, mind you one of them was fudzilla so ... yeah there is always that.

 

[spentshells]

There still does not seem to be any good card to purchase in the $220-300 range. 380X just not worth it. 960 and 380 are below $200 now. GTX 970 and 390 are above $200.

I was hoping for more out of the 380X as well. It still leaves a huge performance and money gap. NVidia can really capitalize with a 960 Ti if they do it right.

yeah especially if the two models rumor is true, pop one on both sides of the 380x with corresponding performance.

 

[superhiperpolulo]

how is 970 faster than the r9 390 when the 390 is supposed to be fasteR?

 

[RedJaron]

I don't doubt you but I read it a few different places, mind you one of them was fudzilla so ... yeah there is always that.

Let's not get into that again.

 

[spentshells]

I don't doubt you but I read it a few different places, mind you one of them was fudzilla so ... yeah there is always that.

Let's not get into that again.

Did this happen before? Do you not believe that diagram I linked to be true? Also I am not seeing any links which state otherwise, I would happily read them.

 

[FormatC]

You would do well to look how sleek Tech power up's charts look like and how neatly per game and overall they are organized.
More to the point of this post- you say that 380x uses virtualy the same idle power with two monitors, while TPU says its typical AMD crapscore of what it used to be- 3-4x more than it should and than you say it is. Did they make a mistake or did you?

TPUs setup seems a little bit unrealistic, if you take a look at the used monitors. I'm using always the same refresh rate for all displays and not monitors from museum, because it reflects the usage behavior a little bit better. And the card is too slow for a real Eyefinity system (x3 or more). Video playback with such a gamer card is also more or less stupid. This card is nothing for HTPCs.

That's what I asked, I don't see why they used a Powercolor 390 and an MSI GTX 970. I don't know, but it just seems a bit unfair and possibly biased toward Nvidia. There is clearly an MSI 390 with the Twin FrozrV coolers just like the 970 that both should have been used for a direct comparison.

I must use, what the companies sent to me. If AMD means it is good enough, it is not my decision. And between you and me: the PCS+ is not slower than the MSI card, only hotter and noisier. And the GTX 970 from MSI is not the fastest on the market. I hate all this discussions about this "Is he biased?". But I can tell you a secret! I'm really biased: I like beef, not pork. 😀

 

[ErikVinoya]

I was hoping for more out of the 380X as well. It still leaves a huge performance and money gap. NVidia can really capitalize with a 960 Ti if they do it right.

I'm really curious as to what NVidia can pull to bring out a 960Ti, or anything between the 960 and 970. The GTX 960 is already using a full GM206 chip, and cutting down a GM204 chip more probably means they would be limited to a 3GB card (unless they're looking for another PR hell).

 

[LuxZg]

Few comments regarding article itself...

First of, you should mark GTX 960 as 2GB in all graphs and in an introductory list of tested cards. It makes GTX 960 look worse because all other cards in nearby range are 4GB cards (380, 380X, 970). Instead, you mention it in passing 1440p testing page and only in Thief section ("Nvidia’s GeForce GTX 960 doesn’t stand a chance, likely due to its 2GB of GDDR5. Unfortunately, we didn’t have a model on-hand with more graphics memory."), so for most people looking for 1080p purchase, and not reading every single word of review - it seems like all test cards are 4GB. And GTX 960 is out of competition. To make matters worse, nVidia is phasing out 2GB models, and is pushing 4GB ones, in clear pursuit of 4GB AMD cards. And we all know that 4GB cards perform much better. I'm complaining about this because I'm looking for a purchasing decision myself, in this very range, and have already noticed that GTX 960 4GB is actually cheaper and quite competitive to AMD-s R9 380 4GB cards (as you say subjective gaming should seem the same on these cards). And this review doesn't do a single thing to draw the comparison between these direct competitors... except if GTX 960 4GB is subjectively same as R9 380, and these are in turn subjectively indistinguishable from R9 380X than probably all three would pass the subjective gaming tests with similar results.

Next up, you extensively test power consumption, temperatures, and noise, yet - we get no information how it competes to other cards (in particular R9 380 and GTX 960). And in the closed case testing of noise you have a probable typo of saying that noise is >1dB above ambient *temperature* .. should be ambient noise I suppose (see table heading row).

All in all, technically great test but as an author and as a publication - you've failed to make a good article which would be based on all this data that you've gathered with your fancy equipment. So in turn, you can just throw all that equipment away, as it means almost nothing to your readers in this form. There isn't even a simple paragraph or two saying something like "temperatures and noise are similar between R9 380 and 380X, GTX 960 requires 20W less power and results in 2dB less noise, while higher end R9 390 and GTX 970 are in range of their own, while again GTX 970 produces better performance with less noise but more power, etc etc". Now THAT would be a conclusion, instead I've got to go read few more reviews to draw the picture myself.

Hopefully you'll learn from these rookie mistakes, and in next review you'll use this equipment to test all cards mentioned (or at least direct competitors), and show a side-by-side results of whatever you test next and it's competitors. Tom's should be in a top few IT hardware reviewing sites, act like it.

 

[LuxZg]

Seems like a good card. Its nice that it fills out the gap between the 960/380 and 979/390. If you look At the BF4 graph, its clear that there is a big difference between these two tiers. If NVIDIA released a 960 ti with 1280 cudas and 4gb ram, we'd have the gap filled out nicely.

Its a great example of how the graphics card market changes over time. 4 years ago a powerful card was released for $500, and now you can buy a very similar card for $240, its slightly more powerful, has more VRAM, and is more efficient.

If this review added results of GTX 960 with 4GB RAM and not 2GB (see my previous comment), specially if they added one with factory OC (as both AMD cards are OCed), than we'd actually know how nVidia competes and what we could expect from eventual/potential GTX 960 Ti 4GB card. To answer that... seems we have to read someone else's review

 

[FormatC]

First of, you should mark GTX 960 as 2GB in all graphs and in an introductory list of tested cards. It makes GTX 960 look worse because all other cards in nearby range are 4GB cards (380, 380X, 970).

In Full-HD it makes nearly no difference if you use the 4GB or 2GB Version - the 380 non-X is mostly faster. I'm waiting for a 4GB model since weeks. The GTX 960 in my test is a card with a higher factory OC. Especially the boost clocks are really high, because it is a hand-picked Golden Sample. In such cases the 2GB model is faster than 4GB model, because the very small memory interface is the same and limiting. But the 4GB model was in rotation, so I had to use the 2GB card from my archive for this review. All benchmarks are fresh with the latest drivers and not refurbished like on some other sites. And the comparison with the other cards... I had only 3 and a half day to benchmark all and write this complete review, graphics and product pictures included! All other cards were described in my single card reviews - too bad, that not one was translated.

Next up, you extensively test power consumption, temperatures, and noise, yet - we get no information how it competes to other cards (in particular R9 380 and GTX 960). And in the closed case testing of noise you have a probable typo of saying that noise is >1dB above ambient *temperature* .. should be ambient noise I suppose (see table heading row).

Bad translation from German. I wrote Delta in original, not Temperature. That means the difference between the system noise (tested with with a passive card) and the whole system with the R9 380X inside.

To be honest:
To compare R9 380, 380X, 390, GTX 960 and 970 in noise and temp ist a little bit more than cherry-picking. This are all custom designs and if I say that a 390 PCS+ from Powercolor is noisier than a R9 380 Gaming 4G from MSI, it is like a comparison between apples and oranges. Other cooler, other power consumption/heat and so on. You can compare cards with the same chip from different brands... This is ok and makes sense.

 

[Little_Gamer]

MR.Author, are you crazy? How 40 fps is not playable? I was laughing. How the heck are you gonna prove that to a mere GMA 3100 (yes I still use it, at least for the next week until I get my 630 - yes , did that too, coz I am a poor person) user -that has a 1280x1024, 2.5GB RAM and a E7500 CPU? Even GTA San Andreas (Jesus Christ) and Team Fortress doesn't run good (TF2 - chris' ma fps , GTA SA MEDIUM - LOW ).

 

[blazorthon]

I'm really curious as to what NVidia can pull to bring out a 960Ti, or anything between the 960 and 970. The GTX 960 is already using a full GM206 chip, and cutting down a GM204 chip more probably means they would be limited to a 3GB card (unless they're looking for another PR hell).

I would be ok with them making a 192 bit / 3GB card between the 960 and the 970. That would fit the bill perfectly, but I doubt they'll do it.

 

[FormatC]

I would be ok with them making a 192 bit / 3GB card between the 960 and the 970. That would fit the bill perfectly, but I doubt they'll do it.

I can guarantee that this card will not come. 3GB are too less for marketing and 6GB are too expensive. The yield rate of all this GM204 ist too good to cripple this chip again, as long as the 970 (and 960) sold so well.

MR.Author, are you crazy? How 40 fps is not playable? I was laughing. How the heck are you gonna prove that to a mere GMA 3100 (yes I still use it, at least for the next week until I get my 630 - yes , did that too, coz I am a poor person) user -that has a 1280x1024, 2.5GB RAM and a E7500 CPU? Even GTA San Andreas (Jesus Christ) and Team Fortress doesn't run good (TF2 - chris' ma fps , GTA SA MEDIUM - LOW ).

For people like you we made also the frame time analysis in three different versions. Try to play f.e. a first person shooter or something similar with 40 fps or less if the frame times are not smooth enough. For some persons values of 25 fps are "playable" but this is nothing for ambitious players. 😉

I've benchmarked Anno 2205 and Fallout 4 with 35 different graphic solutions (VGA cards, APU, iGP) - also for peoples like you with lower settings to get also a R7 240 to run. Simply make a difference between entry level cards and mid-class cards like the R9 380X. AMD has positioned this card as the new 1440p option and not as SVGA card.