XFX R9 380X Double Dissipation XXX OC 4GB Review @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,620
XFX R9 380X Double Dissipation XXX OC 4GB Review - On our test bench today is the XFX R9 380X Double Dissipation XXX OC 4GB video card. It features the latest Ghost Thermal 3.0 cooling technology from XFX and a factory overclock. We will compare it to the ASUS STRIX GTX 960 DCU II OC 4GB in a battle of the $229 price point video cards to determine the better overall value.
 
Looks like they are taking queues from Gigabyte with the voltage locking.

Some people have been resorting to BIOS modding to tweak the voltage. Don't know how far [H] would go for an overclock. lol
 
Wot no rating?

Sad it doesnt allow voltage modding.
I guess its so they were able to coin the phrase "ghost thermal technology".
It only allows for ghost overclocks.
 
Yet another card that shows how badly NVidia needs to step up their game in the midrange. The 960 is overpriced and underpowered. Once upon a time, they corrected that situation by adding a 660ti to offset the shortcomings of the 660. Surely they are smart enough to see the need for a repeat.
 
This card is presently en route to me. Purchased it specifically for Fallout 4. Used Newegg giftcards from Christmas.
 
I also have one of these coming, but mine is an R9-380X-4255. It has the exact same specs as the R9-380X-4DFG, the only difference I can find is the -4255 card says it requires a 650W PSU rather than a 450W.

Anyone know what the difference is?
 
Mugato said:
1080 Please.
Click to expand...

This 2560x1440 benchmark article shows the lowest avg fps score is 43 fps. Either card should be 60+ fps avg at 1080 with these same settings, or even higher quality settings. (Cue posters that want to run everything at max ultra settings on entry level cards.)

I'd like to try a 380X card, but I'm also trying to avoid collecting yet more video cards, lol.

LigTasm said:
I also have one of these coming, but mine is an R9-380X-4255. It has the exact same specs as the R9-380X-4DFG, the only difference I can find is the -4255 card says it requires a 650W PSU rather than a 450W.

Anyone know what the difference is?
Click to expand...

EVGA does this too ... identical specs down to the last number, but two SKUs. My best guess is peripheral changes like box or included swag / adapters. Maybe bios revision?
 
xorbe said:
This 2560x1440 benchmark article shows the lowest avg fps score is 43 fps. Either card should be 60+ fps avg at 1080 with these same settings, or even higher quality settings. (Cue posters that want to run everything at max ultra settings on entry level cards.)

I'd like to try a 380X card, but I'm also trying to avoid collecting yet more video cards, lol.



EVGA does this too ... identical specs down to the last number, but two SKUs. My best guess is peripheral changes like box or included swag / adapters. Maybe bios revision?
Click to expand...

Should be...but you don't know for sure now do you. I would not consider this an "entry" level card, it sit's above the R7 360, R9 370 and R9 380. Below the Fury X, Fury Nano, Fury, R9 390 and 390X. Entry level is 370 and below, this is mid-range or better. As we are talking about 1440p med/high settings. You do realize around 2% of users are running a 1440p monitor? Just a side note. And yes, I do realize we are on the [H] forums.

What's wrong with wanting to max out a game on a budget? Personally, I would rather run max settings at 1080P than run 2560. You get some MSAA oooooo. The game can be much more immersive turning up the shadows and shaders. Or HBAO+. Or GameWorks Fur.
 
The 380x Tonga by itself is pretty good by itself, but I think the upcoming 4GB 390 will push the price down even further.

The XFX 380X DD XtremeXtremeXtreme!!!! card was a pathetic overclocker. A whopping 5% on the core. Memory was equally dismal. Wow, they managed to get 6 X's in the name of this card. that has to be a record.

Both of my used 7970 cards could hit 1200 Mhz on the core F@H with ease. Memory could hit 6 GHZ easy. I had no need to go further.

When it comes to overclocking, Tahiti XT owns Tonga XT.
 
Nihilus1 said:
The 380x Tonga by itself is pretty good by itself, but I think the upcoming 4GB 390 will push the price down even further.

The XFX 380X DD XtremeXtremeXtreme!!!! card was a pathetic overclocker. A whopping 5% on the core. Memory was equally dismal. Wow, they managed to get 6 X's in the name of this card. that has to be a record.

Both of my used 7970 cards could hit 1200 Mhz on the core F@H with ease. Memory could hit 6 GHZ easy. I had no need to go further.

When it comes to overclocking, Tahiti XT owns Tonga XT.
Click to expand...

390 4GB will never see the light of day here, its a china only model.

Tahiti does scale a little better than Tonga, but it just doesn't have the features any more. Freesync is a big one for me, none of the GCN 1.0 chips have it.
 
Mugato said:
Should be...but you don't know for sure now do you.
Click to expand...

Right, that's why I said "best guess, maybe." If that's what you want me to say explicitly, I don't know for sure. I am sure LigTasm would appreciate a concrete answer. There are actually 4 identical part numbers according to TPU database: R9-380X-4255, R9-380X-4DF5, R9-380X-4DFG, R9-380X-F24M. That's as far as I got.
 
xorbe said:
Right, that's why I said "best guess, maybe." If that's what you want me to say explicitly, I don't know for sure. I am sure LigTasm would appreciate a concrete answer. There are actually 4 identical part numbers according to TPU database: R9-380X-4255, R9-380X-4DF5, R9-380X-4DFG, R9-380X-F24M. That's as far as I got.
Click to expand...

Even more interesting, despite what [H] saw in the review all four cards say this on their respective pages:

OTHER FEATURES

Highlight Features - 1 : Double Dissipation

Highlight Features - 2 : Unlocked Voltage
Click to expand...


Somebody isn't telling the truth and I doubt its [H] ;)
 
I have to ask again, why not 280X compared in the review?
 
Araxie said:
I have to ask again, why not 280X compared in the review?
Click to expand...

I think the 380X is "replacing" the 280X in the lineup, thus AMD is not making Tahiti chips any more and once inventory clears you won't be able to buy new ones. At least thats how I understood it, the 380 removed the 280 from shelves as well, very few of them are floating around retail channels still.

Anyway, it doesn't make sense to compare a new card to one that won't be sold any more in a review like this rather than its direct competition which is the 960.
 
LigTasm said:
I think the 380X is "replacing" the 280X in the lineup, thus AMD is not making Tahiti chips any more and once inventory clears you won't be able to buy new ones. At least thats how I understood it, the 380 removed the 280 from shelves as well, very few of them are floating around retail channels still.

Anyway, it doesn't make sense to compare a new card to one that won't be sold any more in a review like this rather than its direct competition which is the 960.
Click to expand...

The R9 285 "replaced" the R9 280 and they were reviewed both to see how much improvement both in performance and power was achieved, the R9 280 resulted a better performer in most games versus the newer R9 285 due to the wider bus with no power consumption advancement in the R9 285 part, so it was a Total Failure... but the 380X, if its really replacing the 280X why isn't reviewed this time? :mad: this card should be reviewed against their direct predecessor to see how much "improvement" was achieved, how much the improvisations do Tonga versus Tahiti in new games, I think, a 380X versus 280X review should be made to see direct comparison from this site instead to have to resort to other sites which did this kind of review.
 
The XFX R9 380X DD XXX OC 4GB is $229.99. We feel the ASUS STRIX GTX 960 DCU II OC is a better deal than the XFX R9 380X DD XXX OC 4GB in this situation, which is currently going for $214.99 after a $15 rebate.
Click to expand...

I thought you guys didn't count rebates when comparing pricing? Oh well, even counting it do you really think the 960 is better value at $15.00 less?
 
xorbe said:
Right, that's why I said "best guess, maybe." If that's what you want me to say explicitly, I don't know for sure. I am sure LigTasm would appreciate a concrete answer. There are actually 4 identical part numbers according to TPU database: R9-380X-4255, R9-380X-4DF5, R9-380X-4DFG, R9-380X-F24M. That's as far as I got.
Click to expand...

I was referring to y our comment about 1440p running at 43fps avg, so 1080P should run at the same or better settings at 60fps avg. Of course it should, but can you max the game out? What settings must be deferred? [H] themselves said this is not a card for 1440, and proved it during the debut of the card. It's not. If you are one of the 2.5% of gamers out there that own a 1440p monitor, chances are you are not going to buy this damn card. So, who is it for? The...1080P crowd!
 
xorbe said:
Right, that's why I said "best guess, maybe." If that's what you want me to say explicitly, I don't know for sure. I am sure LigTasm would appreciate a concrete answer. There are actually 4 identical part numbers according to TPU database: R9-380X-4255, R9-380X-4DF5, R9-380X-4DFG, R9-380X-F24M. That's as far as I got.
Click to expand...

We asked them about the part number spaghetti when I did a review of their 390X card, and XFX stated that the part numbers were more for different regions that the cards were sold in. Though, when I was looking around, I saw 2 of the 4 part numbers readily available in the US with the same specs... So, you can take the source's word on it or make your own guess about what it is :)
 
Or buy a different mfrs card when you cant be sure what you are getting.
 
xorbe said:
This 2560x1440 benchmark article shows the lowest avg fps score is 43 fps. Either card should be 60+ fps avg at 1080 with these same settings, or even higher quality settings. (Cue posters that want to run everything at max ultra settings on entry level cards.)

I'd like to try a 380X card, but I'm also trying to avoid collecting yet more video cards, lol.



EVGA does this too ... identical specs down to the last number, but two SKUs. My best guess is peripheral changes like box or included swag / adapters. Maybe bios revision?
Click to expand...

Something I've noticed, they use special sku's to avoid rebates or price matching at best buy.
 
Nihilus1 said:
I would like to see it compared to an HD 7970 so we can see what is changed in 4 years time.
Click to expand...

There shouldn't be too much difference between the two, other than 7970 level performance is now roughly mid tier level performance.

Still pretty good 4 years later.
 
Nihilus1 said:
I would like to see it compared to an HD 7970 so we can see what is changed in 4 years time.
Click to expand...

The main difference is feature set and power usage. There's additional performance in certain metrics (tessellation and pixel fill rate, for example) but it might or might not show in the games [H] benches.

78749.png


78751.png
 
Relayer said:
I thought you guys didn't count rebates when comparing pricing? Oh well, even counting it do you really think the 960 is better value at $15.00 less?
Click to expand...

Hardocp gives a new meaning to unbiased.
 
Relayer said:
I thought you guys didn't count rebates when comparing pricing? Oh well, even counting it do you really think the 960 is better value at $15.00 less?
Click to expand...

Remon said:
Hardocp gives a new meaning to unbiased.
Click to expand...

It depends, when the cards are so small or low in pricing, a rebate can make a big difference when talking about cards in the $200 range. More expensive cards are less sensitive to rebate offerings.

Naturally, rebate prices do change and are dynamic, so you shouldn't bet on those prices remaining. However, we wouldn't be a good resource of information if we did not relate good deals at online stores to you guys to lower the price of purchase that is available. Surely you want to know about current good deals on hardware.

Which is a better value is subjective, if I had to chose personally I'd spend a little more money for a high factory overclocked 380X 4GB, The ASUS 960 used in this review is a highly clocked factory card. It also has a higher overclockability potential currently compared to this card that we couldn't unlock voltage control. So those things combined, does give more value to the 960.

Overall, this particular XFX 380X card had a very tame factory overclock, and we've seen better, certainly the ASUS STRIX 380X has 1030MHz out of box clock, which makes more sense on the 380X. This 960 used offers a better factory overclock, and we'd easily suggest you go with a higher factory overclock on the 380X if you are going to get one, trust me, it needs that higher clock speed to compete well. With the headroom on 380X this particular XFX card has a low factory oc and tame oc potential.

We are trying to lineup more 380X cards, ones with higher overclocks and custom cooling, which are appealing for this GPU. If XFX ever offers a higher clocked 380X, we will of course request it as well.
 
Last edited:
Relayer said:
The main difference is feature set and power usage. There's additional performance in certain metrics (tessellation and pixel fill rate, for example) but it might or might not show in the games [H] benches.
Click to expand...

Thanks! After going to that review, the 380x looks rather disappointing.

Yakk said:
There shouldn't be too much difference between the two, other than 7970 level performance is now roughly mid tier level performance.

Still pretty good 4 years later.
Click to expand...

On the game benchmarks, the 7970 was neck and neck with the 380x except for ME:SOM and Total War. This is probably because of the vRAM size since I know Shadows of Mordor is a vRAM hog.

The 7970 consumes only 15 watts more, overclocks a HELL of a lot better, and can be found for very cheap.

I would say it held up VERY well after 4 years.
 
Well I got my -4255 card in, it is voltage locked as well. Also, its pretty damn loud.

EDIT: I just noticed it runs at 1040/5800 instead of the advertised 990/5700.
 
Last edited:
Nihilus1 said:
On the game benchmarks, the 7970 was neck and neck with the 380x except for ME:SOM and Total War. This is probably because of the vRAM size since I know Shadows of Mordor is a vRAM hog.

The 7970 consumes only 15 watts more, overclocks a HELL of a lot better, and can be found for very cheap.

I would say it held up VERY well after 4 years.
Click to expand...

Yup, all the while leaving it's original competition behind. People can complain of rebrands, but when whatever new flavour is released doesn't bring much to the table I rather get a rebrand with continued *active* product support for years, not just "check box" support like we can see so often in other products which doesn't improve anything past the first year.

Also we can see why AMD wasn't in a big hurry to release a full Tonga card, especially if they possibly had a lot of cut down chips to sell from their supplying Apple iMacs.
 
I think you guys got a lemon of a XFX 380x DD card. I've already got mine to 1060mhz w a +10 bump in power and have not even tried to OC it yet. I'm using the Asus OC utility. got 7800 in 3dmark w/ a non-oc'd i5-6400 w/ just the bump in core on the card. No artifacting on my card.
 
Up to 1100/6100 now, no artifacting, 7881 in 3dmark w/ Intel i5-6400 (CPU not oc'd)
 
man, overall 3Dmark score are useless, if you want to post anything of progress, post Graphic score only..
 
Graphic Score only, like what would you want me to post? Graphic Score.
 
Well I do know with this card, that I got at a great deal on Amazon, 229.00 w/ 20$ rebate that I at least can run Fallout 4 in 1920x1080 w/ all features maxed out. I think I got a good deal at least.
 
Evil said:
Graphic Score only, like what would you want me to post? Graphic Score.
Click to expand...

look, 3dmark throw 4 results in fire strike.

one Overall the biggie one. one Graphic score, one Physics score and one Combined score. Physics score means nothing as it vary from CPU to CPU and show day to night differences for example from i5 to i7 4c/8t. combined score is a mix from physics + Graphics test generally limited by CPU again useless to compare GPUs. Overall results i'ts a combination of all, graphics, physics and combined test results, this again mean nothing.

The result you have to pay attention its the one called Graphics score.. that's the one that really matter to compare GPUs. just as example i'll post my last result tested with my old 280X. im not using that gpu anymore so bear with me for old results.

FireStrike:
Overall Score: 8069
Graphic Score: 9096
Physic Score: 12087
combined Score: 3441


3dMark11
Overall Score: 11439
Graphic Score: 11802
Physic Score: 11007
Combined: 9762
 
Evil said:
Using latest AMD Crimson driver, yet 3dmark reports not approved
Click to expand...

it does that most of the time. Never likes beta, although rarely it gets by. And whql releases should always be approved but sometimes register as unapproved. Really it means nothing.
 
OmegaGame said:
You guys did say you would try the strixx OC utility to see if it offers unlocked voltage control.. Yet it would seem you never did, Because it does have voltage control with strixx and i was able to push mine to 1170 mhz core before i noticed any artifacts all while staying ice cold.
Click to expand...

As of 1/8/2016 we tested it with TRIXX and it did not work with the current software at the time of review. This was a few days after the article was published, we checked it with TRIXX, therefore there was no follow-up article we could do.
 
You must log in or register to reply here.
Back
Top