XFX R9 290X Black OC Edition CrosFire Video Card Review @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
XFX R9 290X Black OC Edition CrosFire Video Card Review - Today we review two XFX R9 290X Black OC Edition video cards in CrossFire to see what quality 4K gaming we can achieve. We will be comparing this to a SLI GeForce GTX 970 configuration, in hopes of clearly defining the better overall value in this high-end gaming category. All for $699.98 after $60 MIR.
 
I had to log-in just to comment on this review.

Once you saw something was seriously off with the OC results, you should've canceled this review or search for issue why this happens.

290X in Crossfire at 4K are as fast or faster usually than SLI 980s. There is no way in hell Xfire of 290Xs is some 15-20% slower than SLI 970s.

OC'd 290Xs in 4K should be able wipe the floor with 970s, no questions asked.


PS: In Witcher 3, Crossfire, enabling AA currently breaks Crossfire. It's in the driver release notes. Disable AA to proper compare.
 
I just got this card at 270$ new and guess I will not be crossfireing it. My plan is to stick to 1440p with freesync until the 400 series comes out with enough RAM to handle 3x1440p.

Thanks for the review and warnings.
 
Nice review, but this is confusing. I've seen much different results from other 290x crossfire/ 295x reviews. They seem to scale rather well. Even though the 290x is considerably slower than the GTX 980, it often keeps up when it is in crossfire vs. SLI - yet this review shows it getting its teeth kicked in by 970 SLI.
 
I had to log-in just to comment on this review.

Once you saw something was seriously off with the OC results, you should've canceled this review or search for issue why this happens.

290X in Crossfire at 4K are as fast or faster usually than SLI 980s. There is no way in hell Xfire of 290Xs is some 15-20% slower than SLI 970s.

OC'd 290Xs in 4K should be able wipe the floor with 970s, no questions asked.


PS: In Witcher 3, Crossfire, enabling AA currently breaks Crossfire. It's in the driver release notes. Disable AA to proper compare.
Their graphics do show that AA was disabled :confused:.

You mean Apples-to-apples? It doesn't look too much slower than their test with AA off.

Reading compreshension... A2A is at 1440p, not 4k...
 
*Grabs pitchfork* Yes Kyle, cancel every review you guys work on until Amd wipes the floor with Nvidia! The fans demand it! /sarcasm

Thanks for the results! Even with the good results from the oc 290X's they would have still been behind the 970s
 
Did you try CrossFire with non-XFX 290Xs to see if there was any difference? I never had that experience with my 290X cards from Sapphire. That's very odd.

I know you guys are professionals so I doubt this is the case but you did check to make sure you were not throttling in CFX right?
 
First of all I have the same bug where if I OC the core on my R9 290, I get significantly less performance. But I have other issues with a new Win 8.1 install so I'm not going to blame AMD drivers. Every time that I switch sound from Headphones to Speakers and vice versa on my SoundBlaster Zx, I have to CTRL-ALT-DEL and end the task on it and every program that uses sound on my PC. Like I said this is a new WIn 8.1 install and I have tried to install it 3 times. I didn't have a single issue with this before I reinstalled Win 8.1. I'm just giving up for now and waiting for the 29th of July. Everything works but OC of my video card and switch the sound output. That's not too bad. It is annoying though. :mad:

Secondly going by these numbers I wouldn't run SLi or CorssfireX on these cards. In every new game the performance was below 60fps even with so many options turned off. That means that you don't have enough horsepower. It's nice that they can run older titles at 4K though which is something that I'm really into. Thank you for testing that! But being able to run older titles at 4K isn't going to help you with next month's AAA game though when you try for 4K. These cards need to stay at 1440p resolutions.

Perfect example of this is the Advanced Blending option in DiRT Rally. It is more of a frame rate killer than GTA V's grass options and view distance combined. It adds a lot of beauty to the forest levels and grass though. But 30fps gaming is just not fun to me.

Thanks again for the review. If you figure out what is breaking the OC please tell us!
 
*Grabs pitchfork* Yes Kyle, cancel every review you guys work on until Amd wipes the floor with Nvidia! The fans demand it! /sarcasm


Yes, in future Kyle please just disable any settings that make the AMD card look bad or that the drivers can't cope with and then base the "apples to apples" off the bad apple. :)

I mean, who really cares about AA anyway, amirite?
 
Would have been interesting to see a 980 Ti included in the review that had been overclocked as well. Think you'd find that a overclocked 980Ti can keep up with 970SLI/290X Crossfire.
 
Lolno, 980 Ti is like 15-20% slower than Crossfire where Crossfire really works.
 
I had to log-in just to comment on this review.

Once you saw something was seriously off with the OC results, you should've canceled this review or search for issue why this happens.

290X in Crossfire at 4K are as fast or faster usually than SLI 980s. There is no way in hell Xfire of 290Xs is some 15-20% slower than SLI 970s.

OC'd 290Xs in 4K should be able wipe the floor with 970s, no questions asked.


PS: In Witcher 3, Crossfire, enabling AA currently breaks Crossfire. It's in the driver release notes. Disable AA to proper compare.

Explain to me why they need to cancel their review? It's not their job to jury rig things so the product works. It either does or it doesn't, and in this case it didn't. Suck it up. That's what reviews are for.

Would you have been so adamant if it were the 970 that couldn't overclock? I doubt it.

Thanks for the review guys! I sure hope you get the overclocking issue figured out on these things so there can be some kind of update. However, since we are all pretty familiar with Crossfire 290x performance already there probably isn't a lot of value in that.
 
Explain to me why they need to cancel their review? It's not their job to jury rig things so the product works. It either does or it doesn't, and in this case it didn't. Suck it up. That's what reviews are for.

Would you have been so adamant if it were the 970 that couldn't overclock? I doubt it.

Thanks for the review guys! I sure hope you get the overclocking issue figured out on these things so there can be some kind of update. However, since we are all pretty familiar with Crossfire 290x performance already there probably isn't a lot of value in that.

Lots of value in knowing if you should upgrade to something new and what newer drivers bring to the table. Remember AMD isn't Nvidia. AMD keeps the same basic GCN architecture going from generation to generation. So next year's driver improvements will likely benefit R9 290X users just as much as whatever is released on a new process node.
 
I'm more baffled on why a review would be posted on a dead architecture.
 
Since the cards are already overclocked, did you try running at "stock" and see if the issue persisted?
 
I'm more baffled on why a review would be posted on a dead architecture.
Because people usually decide on what to buy based on the price to performance of what's available in the market today. The 290X cards that are still available compare favorably to the 970 at similar price points. Hawaii is still being sold as the 300 series so it's far from dead.
 
Explain to me why they need to cancel their review? It's not their job to jury rig things so the product works. It either does or it doesn't, and in this case it didn't. Suck it up. That's what reviews are for.

Well if one of the 290x's was defective I think that would warrant stopping the review. Doesnt make much sense to review a pair of good working 970's against a pair of 290x's where would looks to be defective or something was up. Im not saying thats what happened here but I can see why it raises a flag or two for people. I do think I wouldnt have included the OC'd 970's if it were me.

These results look about right judging by other reviews Ive seen so I dont think its a broke card. Its only The Witcher and GTA where the 290x's fall short, and theyre winning in the other games. Looking around other sites thats about how it goes. If one card was broke, the results would be more skewed. However the near 50% loss in The Witcher makes me scratch my head. That doesnt seem to scale with the other results. But since AMD hasnt released a new driver since the release of The Witcher except for a Beta, that could be the issue.

Overall I consider them about even as theyre trading wins in about the same number of games.
 
Is there a bug with GTA V on AMD video cards? I have been getting the same fluctuations playing at 1080p and then eventually I experience a lot of stuttering on my R9 290, even with the settings as low as possible. I dusted off my GTX 480 and it plays smoothly with a lot of the settings on High. I started reading some threads on the Rockstar forums that suggested the Beta Catalyst driver, but that has not helped. Your GTA V results were all over the place too.

It doesn't seem right that the game plays better on my Alienware Alpha (i3 & GTX 750 Ti) than on my i7-2600K and R9 290.
 
Because people usually decide on what to buy based on the price to performance of what's available in the market today. The 290X cards that are still available compare favorably to the 970 at similar price points. Hawaii is still being sold as the 300 series so it's far from dead.

The 290 series supply is drying up. And the 390 series had some tweeks as well at 5GHz memory.

To quote the review, "the 390 is just a rebadge of the 290" is a bit misleading as there is twice the memory and a 25% boost in memory clock rate.

So HardOCP is reviewing a <mostly or soon to be> dead product. Then they implied the 390 won't do better.

Just a poor choice of reviews and wording.
 
Kyle, it could be worth trying the overclocks again with a driver pre release of the 390 cards.
I get a feeling that AMD are trying to make the 290s look worse value than the 390s and perhaps they have done something to the latest driver.

A bit extreme thinking but there is reason for my suspicion.
First, the new cards use a driver that isnt compatible with the older cards, this immediately didnt seem right with them being the same architecture.
Then the 390x showed abnormally good performance in some tests compared to the 290x suggesting they havent passed on some improvements to the driver for the 290 cards.
They may have gone a step further.
 
Kyle, it could be worth trying the overclocks again with a driver pre release of the 390 cards.
I get a feeling that AMD are trying to make the 290s look worse value than the 390s and perhaps they have done something to the latest driver.

A bit extreme thinking but there is reason for my suspicion.
First, the new cards use a driver that isnt compatible with the older cards, this immediately didnt seem right with them being the same architecture.
Then the 390x showed abnormally good performance in some tests compared to the 290x suggesting they havent passed on some improvements to the driver for the 290 cards.
They may have gone a step further.

Mine started as soon as the "Register to update to WIndows 10 for free" screen showed up.
 
Mine started as soon as the "Register to update to WIndows 10 for free" screen showed up.

You had a drop in performance when overclocked from then on, or did you mean something else?
 
Kyle, it could be worth trying the overclocks again with a driver pre release of the 390 cards.
I get a feeling that AMD are trying to make the 290s look worse value than the 390s and perhaps they have done something to the latest driver.

A bit extreme thinking but there is reason for my suspicion.
First, the new cards use a driver that isnt compatible with the older cards, this immediately didnt seem right with them being the same architecture.
Then the 390x showed abnormally good performance in some tests compared to the 290x suggesting they havent passed on some improvements to the driver for the 290 cards.
They may have gone a step further.

I think this could be serious, I know that's a typical nvidia move, just stop to support and improve older gen cards in newer game tittles, but AMD doing that? is rare, however possible.. I agree, it may require further investigation.
 
Would have been interesting to see a 980 Ti included in the review that had been overclocked as well. Think you'd find that a overclocked 980Ti can keep up with 970SLI/290X Crossfire.

I was hoping to see the same - I am starting to get some driver errors with my SLI 680s and am about to make an upgrade (I think my original Galaxy 680 which runs really hot is crapping out). With my resolution (1600p) I want something that is going to last a couple of years and I would really prefer to go with a single card to avoid any stuttering issues, but often time it's more economical to go SLI.

With reviews like this, I will find myself going between multiple reviews (aka 980ti/390x/Fury) to get meaningful buying and performance comparisons. It would be nice if you had a compilation page of review results so it would be easier to compare results across different reviews. I say this as I have come to value the methodology here on HARDOCP in my buying decisions and I have not been let down.
 
I had to log-in just to comment on this review.

Once you saw something was seriously off with the OC results, you should've canceled this review or search for issue why this happens.

290X in Crossfire at 4K are as fast or faster usually than SLI 980s. There is no way in hell Xfire of 290Xs is some 15-20% slower than SLI 970s.

OC'd 290Xs in 4K should be able wipe the floor with 970s, no questions asked.


PS: In Witcher 3, Crossfire, enabling AA currently breaks Crossfire. It's in the driver release notes. Disable AA to proper compare.

I game at 2560x1600 and don't use AA, the other sites do this shit too. The Hexus Gigabyte G1 980 ti review was more helpful to me because it included the 7970. There was also a problem in the Fury X review, during the witcher 3 segment when the 290X was run with high settings instead ultra which makes the comparison to the other two cards useless. I wanted to see how well the Fury X did over the 290x at the SAME settings.


Did you try CrossFire with non-XFX 290Xs to see if there was any difference? I never had that experience with my 290X cards from Sapphire. That's very odd.

I know you guys are professionals so I doubt this is the case but you did check to make sure you were not throttling in CFX right?

Yes not including standard 290Xs in this gives us no baseline. This was a lazy review please go back and do the jov the correct way.
 
Did you try CrossFire with non-XFX 290Xs to see if there was any difference? I never had that experience with my 290X cards from Sapphire. That's very odd.

I know you guys are professionals so I doubt this is the case but you did check to make sure you were not throttling in CFX right?

Since the cards are already overclocked, did you try running at "stock" and see if the issue persisted?

Reading is fundamental.
 
So now in reviews we need to specify every thing that we did NOT do as well?

No, we did not downclock the cards. We ran the cards at their stock clock per XFX specifications. No, this review is not about non-XFX cards, we did not retest those cards for this review either. However, there are more than a few that you can look up that we have done since the 290X launched.
 
You had a drop in performance when overclocked from then on, or did you mean something else?

Yes, that was about the same time period that I noticed that my games would run like crap. I bought 3DMark during the summer sale for $5. That's when I discovered that my score would plummet if I OC the core of my card. The memory didn't seem to be affected by the bug.

So I'm not going to say that on the day that Windows 10 signup was introduced it started. But it was right around that time period that I was pulling my hair out trying to figure out why games like Euro Truck Simulator 2 would run at 20 fps instead of 60fps if I OC'd my card.

Hope that helps. :)
 
Yeah strange OC results, certainly not typical.

I wonder if XFX skimped on the VRM cooling so they ended up running hot even at stock speeds. Its not uncommon to see 290X's that have VRM temps 10+ degrees over the core temp even in single-card.
 
I wonder if your PSU is perhaps getting a tad weak, and the voltage (on either 12v or 5v rails, perhaps just the rail running the Video cards) might be dropping a bit when you overclock? The AMD cards are crazy power hungry as it is, and overclocking tends to add quite a bit to the load. Might be worth measuring the voltages during an xfire oc to see what happens.


That is kinda interesting, but the results are a bit hard to directly compare, since the AA and AO options used in each review is different. On the only one I spotted the same AA settings on, going from a 970sli in nov14 at No AO to the oc970sli today with HBAO, and a few fps faster also, is a nice improvement (but its oc vs non-oc).
 
I think you misunderstood what I was asking. I was asking if the cards were potentially throttling when overclocked in CrossFire due to power consumption or heat issues. The cooler on the XFX DD isn't that great and since it dumps the heat in the case I was wondering if the top card, which gets hotter in my experience, was throttling down due to overheating, which wouldn't have been a problem with a single card installed.

We watch the GPU clocks live, while each game is running during overclocking, in this case it was not throttling. We even upped the fan speeds, as noted in the review, to alleviate any throttling due to thermals.

To any other questions r.e. the overclock, I'm going to investigate CF overclocking with these two cards on my system as well, and contact AMD and XFX to get to the bottom of it. Eventually we will be doing 390X CrossFire OC testing, and we will probably use these cards as a comparison, so I will test them and see if anything can be done, if not, we have two other XFX DD cards, non black edition, I will also attempt to overclock in CrossFire and see what happens.
 
I had to log-in just to comment on this review.

Once you saw something was seriously off with the OC results, you should've canceled this review or search for issue why this happens.

290X in Crossfire at 4K are as fast or faster usually than SLI 980s. There is no way in hell Xfire of 290Xs is some 15-20% slower than SLI 970s.

OC'd 290Xs in 4K should be able wipe the floor with 970s, no questions asked.


PS: In Witcher 3, Crossfire, enabling AA currently breaks Crossfire. It's in the driver release notes. Disable AA to proper compare.



lol.. exactly.
 
lol.. exactly.

two members of the forum brain trust here i guess?

so according to your reasoning. as soon as [H] stumbles upon a potentially platform breaking failure, the best course of action is to pretend that it didn't happen.

a few years back when they reported the anomaly where gtx 460 sli was faster than hd 5870 cf, the editors should have kept their traps shut and spared amd any embarrassment. amd rededicated themselves to improving driver support soon after so i suppose that was a good thing.

when [H] chastised amd for its poor hd 7970 multi gpu smoothness, that too was a missed opportunity to hold their tongues and allow amd to save face, right? Then again, amd decided it would double its efforts toward remedying its multi card smoothness. as a result we have the frame pacing algorithm present in amd drivers today and their dma technology employed on their top tier products.
 
two members of the forum brain trust here i guess?

so according to your reasoning. as soon as [H] stumbles upon a potentially platform breaking failure, the best course of action is to pretend that it didn't happen.

a few years back when they reported the anomaly where gtx 460 sli was faster than hd 5870 cf, the editors should have kept their traps shut and spared amd any embarrassment. amd rededicated themselves to improving driver support soon after so i suppose that was a good thing.

when [H] chastised amd for its poor hd 7970 multi gpu smoothness, that too was a missed opportunity to hold their tongues and allow amd to save face, right? Then again, amd decided it would double its efforts toward remedying its multi card smoothness. as a result we have the frame pacing algorithm present in amd drivers today and their dma technology employed on their top tier products.

Yeah but the problem is that these 290X's are performing much worse than a previous pair of 290X's that [H] tested. The GTX 460 SLI thing was an obvious problem with the AMD cards. This is different because these cards performed a whole lot better in an earlier review. I know it's a it a conspiracy or anything like that but there is an awful lot of evidence to make you think something is wrong somewhere.
 
Yeah but the problem is that these 290X's are performing much worse than a previous pair of 290X's that [H] tested. The GTX 460 SLI thing was an obvious problem with the AMD cards. This is different because these cards performed a whole lot better in an earlier review. I know it's a it a conspiracy or anything like that but there is an awful lot of evidence to make you think something is wrong somewhere.

when was the last time this site published an article of 290x cf at 4k? from what i can tell, it hasn't been since the gtx 970 4k review 7 months ago. How can you even compare the two sets of results when the author, test systems, and driver versions are different? there are only two games that are used in both articles. where's all the evidence to suggest that [H] fudged the results?
 
when was the last time this site published an article of 290x cf at 4k? from what i can tell, it hasn't been since the gtx 970 4k review 7 months ago. How can you even compare the two sets of results when the author, test systems, and driver versions are different? there are only two games that are used in both articles. where's all the evidence to suggest that [H] fudged the results?

Bad week for AMD fanbois as Fury hardly distinguished itself and the power draw of 290x in Crossfire is insane. Nvidia is really putting the hurt on AMD since Maxwell came out and AMD clearly has not caught up.
 
Back
Top