SAPPHIRE VAPOR-X R9 290X TRI-X OC Video Card Review @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,634
SAPPHIRE VAPOR-X R9 290X TRI-X OC Video Card Review - We take a look at the SAPPHIRE Vapor-X R9 290X TRI-X OC video card which has the highest factory overclock we've ever encountered on any AMD R9 290X video card. This video card is feature rich and very fast. We'll overclock it to the highest GPU clocks we've seen yet on R9 290X and compare it to the competition.
 
Jeeeez nice card.

Typo Fixed, thanks for the extra eyes. - Kyle
 
Last edited by a moderator:
did you test it in crossfire and anything strange with UEFI boot behavior when running 3 monitors? like not showing the bios or black screening?

no black screening when OCing or gaming at all?
 
Adidas4275 said:
did you test it in crossfire and anything strange with UEFI boot behavior when running 3 monitors? like not showing the bios or black screening?

no black screening when OCing or gaming at all?
Click to expand...

No CrossFire testing done. No issues other than what is spelled out specifically on the overclocking page.
 
Darksword said:
Nice card.

But for $580 http://www.newegg.com/Product/Product.aspx?Item=N82E16814202046 you can get R9 280X Crossfire, which will outperform this card quite a bit - as long as 3GB VRAM is enough for you.
Click to expand...

Then you have to deal with having 2 videocards in a system. And that extra 1GB of video memory makes a huge difference in Watch Dogs. I suspect games are going to require more memory in the future as more PS4 / XBONE console ports show up. Those machines have larger videocard memory requirements as the Developers working on games have access to more of it on those consoles.

And to Kyle and the crew. Great review as always!
 
close to 700 bills and not even a game bundle. Cheap Sob.
Great looking card but to rich for my blood.
 
Typo on page 2: no such thing as ForceWare 350.52 WHQL, should be Geforce 340.52 WHQL.

Then a serious question. Why is it that [H] systematically refuses to include overclocked GK110 numbers in a direct comparison to an overclocked Hawaii? The only time you've broken this rule is pretty much the only scenario where the overclocking prowess of the GK110 can't help it edge out the Hawaii equivalent (4K multi-GPU). Taking a look at your separate GTX 780 Strix OC performance numbers, even the non-Ti 780 properly overclocked would have put up a decent fight against this top model 290X. Going to such lengths to point out the matching price with the 780Ti and then only overclocking one of the cards... It just seems curious to me.

I'm also still waiting for that full Watch Dogs article. As for my own experience with the game, the basic problem causing the stuttering is not running out of VRAM because it is still there even with lowered settings and VRAM usage not hitting the cap. There is something broken with the texture streaming which seems to affect Nvidia cards more than it does AMD. That being said, at your resolution and settings the 3GB cards really are maxing out their VRAM as well. Double trouble there. It's just that many, many 4GB or even 6GB card owners are reporting really bad stuttering as well... Actually, I haven't seen any difference in the stuttering between my 780 3GB and my 670 4GB GPUs. So it really isn't as simple as "running out of VRAM".
 
specops said:
Typo on page 2: no such thing as ForceWare 350.52 WHQL, should be Geforce 340.52 WHQL.

Then a serious question. Why is it that [H] systematically refuses to include overclocked GK110 numbers in a direct comparison to an overclocked Hawaii? The only time you've broken this rule is pretty much the only scenario where the overclocking prowess of the GK110 can't help it edge out the Hawaii equivalent (4K multi-GPU). Taking a look at your separate GTX 780 Strix OC performance numbers, even the non-Ti 780 properly overclocked would have put up a decent fight against this top model 290X. Going to such lengths to point out the matching price with the 780Ti and then only overclocking one of the cards... It just seems curious to me.

I'm also still waiting for that full Watch Dogs article. As for my own experience with the game, the basic problem causing the stuttering is not running out of VRAM because it is still there even with lowered settings and VRAM usage not hitting the cap. There is something broken with the texture streaming which seems to affect Nvidia cards more than it does AMD. That being said, at your resolution and settings the 3GB cards really are maxing out their VRAM as well. Double trouble there. It's just that many, many 4GB or even 6GB card owners are reporting really bad stuttering as well... Actually, I haven't seen any difference in the stuttering between my 780 3GB and my 670 4GB GPUs. So it really isn't as simple as "running out of VRAM".
Click to expand...

@Specops I thought the MSI GTX 780ti wasn't just an overclocked model, but the best overclocked GTX 780ti on the market? Read the last paragraph here.

At some points during the article a disclaimer said that a normal GTX 780ti couldn't handle the settings that the MSI GTX 780ti because of the higher clock settings on the MSI. That gave me the impression that the MSI GTX 780ti was overclocked. Here is a quote from the Crysis 3 section.

"The factory overclock on the MSI GTX 780 Ti GAMING also allows this game to be playable at "Very High" with SMAA 2TX. Note that a reference GTX 780 Ti cannot run this game at "Very High" like the reference AMD R9 290X."

Seems like the MSI GTX 780ti was overclocked to me.

Interesting thing I found out about Watch Dogs is that overclocking the memory on my R9 290 made the biggest difference. I didn't even overclock it very much to fix many of my stuttering issues.
 
"Factory overclocked" isn't overclocked. Overclocked is the max sustainable clocks. Such as what the 290X Vapor-X was running as its "OC" clocks compared to the "factory overclock" 1080MHz clock rates.
 
specops said:
Typo on page 2: no such thing as ForceWare 350.52 WHQL, should be Geforce 340.52 WHQL.

Then a serious question. Why is it that [H] systematically refuses to include overclocked GK110 numbers in a direct comparison to an overclocked Hawaii? The only time you've broken this rule is pretty much the only scenario where the overclocking prowess of the GK110 can't help it edge out the Hawaii equivalent (4K multi-GPU). Taking a look at your separate GTX 780 Strix OC performance numbers, even the non-Ti 780 properly overclocked would have put up a decent fight against this top model 290X. Going to such lengths to point out the matching price with the 780Ti and then only overclocking one of the cards... It just seems curious to me.

I'm also still waiting for that full Watch Dogs article. As for my own experience with the game, the basic problem causing the stuttering is not running out of VRAM because it is still there even with lowered settings and VRAM usage not hitting the cap. There is something broken with the texture streaming which seems to affect Nvidia cards more than it does AMD. That being said, at your resolution and settings the 3GB cards really are maxing out their VRAM as well. Double trouble there. It's just that many, many 4GB or even 6GB card owners are reporting really bad stuttering as well... Actually, I haven't seen any difference in the stuttering between my 780 3GB and my 670 4GB GPUs. So it really isn't as simple as "running out of VRAM".
Click to expand...

specops said:
"Factory overclocked" isn't overclocked. Overclocked is the max sustainable clocks. Such as what the 290X Vapor-X was running as its "OC" clocks compared to the "factory overclock" 1080MHz clock rates.
Click to expand...

HERE you go.
 
specops said:
"Factory overclocked" isn't overclocked. Overclocked is the max sustainable clocks. Such as what the 290X Vapor-X was running as its "OC" clocks compared to the "factory overclock" 1080MHz clock rates.
Click to expand...

because this is a review of the Sapphire Vapor-X R9 290X dsjdfghkljdfhg card, not the MSI GTX780Ti wriuyertiopu card, if you want the numbers for that go look them up yourself or just click the link relayer posted instead. personally i like the overclock numbers and to see what a cards capable of but i think it gives the less tech savvy bad information on whether one card is better than the other when you start comparing overclocks to each other in a product review.


/slams face into desk, who the hell comes up with these friggin names for graphic cards.
 
Last edited:
Fantastic video card and a fantastic price too. Way overkill for me at 1080p gaming. ;)
 
Pretty stout card.
Awesome feature set.

Worth the money in my book. Two please.

Wonder how it would do if you put a waterblock on it?:eek:
 
That OC is meh. Almost not even worth OC'ing the card from where the manufacturer set it to.
 
sinnathurai said:
Is it the 8gb version
Click to expand...

This says it all: Today we are interested in the Vapor-X technology that is featured on the SAPPHIRE VAPOR-X R9 290X 4GB GDDR5 TRI-X OC (UEFI) video card we are evaluating today.
 
specops said:
"Factory overclocked" isn't overclocked. Overclocked is the max sustainable clocks. Such as what the 290X Vapor-X was running as its "OC" clocks compared to the "factory overclock" 1080MHz clock rates.
Click to expand...

AS noted on Page 4, at the start of testing "Note 2 - The MSI GTX 780 Ti GAMING video card is running in OC Mode. The actual in-game frequency is a high 1162MHz in all our gameplay tests.

JaiWebb said:
That OC is meh. Almost not even worth OC'ing the card from where the manufacturer set it to.
Click to expand...

While I know I "only" have a 290, my launch Sapphire (Elpida ram at that) running the Tri-X bios runs @ 1000/1300 with a voltage of 1.16V..With my max OC I am at 1275/1375 @ 1.31V (can run as low as 1.25V but will crash after 5 or 6 hours..The 1.31V is stable under a 24Hr loop of Heaven with max setting and the CPU load 100% as well)..

I just assumed that was a decent but nothing speical OC, but apparently I was wrong in thinking the 290/290X cards clock higher then 1150+ on average..I have really lucked out when it comes to very high clocking GPUs, my last card, the MSI 7950 TF3 did 1.325Ghz core and 1.750Ghz memory 100% stable, although I run all my cards under water with a very beefy cooling loop.

IF I was one to stick to air cooling, and wanted a 290X, I think I would pony up the extra cash for this guy. As it stands, you can a new 290 and slap on a full cover block and most of a decent WC loop for the price of this card alone..I think going the 290+WC'ing route is a no brainer since you can also add your CPU into the loop for as little as ~$45 for a decent CPU block (less if you buy used parts)..
 
Since so many of you are eager to point out: yes, the tested 780Ti is a custom factory overclocked card. Now, get my point: it is not running max clocks but rather the factory standard clocks. It was not overclocked by [H] whereas the Vapor-X was. See? Two factory overclocked cards, only one of them further overclocked to max stable clocks. Not consistent.

I know [H] as well as many other hardware sites usually make comparisons to stock cards only. I'm just saying that it is a bit curious to see [H] actually go and overclock both a GK110 and a Hawaii but only on the 4K CF/SLI review. I would have loved to see a similar effort with single GPU and "normal" resolution as well.
 
specops said:
Since so many of you are eager to point out: yes, the tested 780Ti is a custom factory overclocked card. Now, get my point: it is not running max clocks but rather the factory standard clocks. It was not overclocked by [H] whereas the Vapor-X was. See? Two factory overclocked cards, only one of them further overclocked to max stable clocks. Not consistent.

I know [H] as well as many other hardware sites usually make comparisons to stock cards only. I'm just saying that it is a bit curious to see [H] actually go and overclock both a GK110 and a Hawaii but only on the 4K CF/SLI review. I would have loved to see a similar effort with single GPU and "normal" resolution as well.
Click to expand...

if you read the [H] review of the the msi card in this review, you'd know that they only managed to squeeze out and additional 25mhz which would translate to almost nothing in terms of improvement over simply running the card in its factory spec oc'd mode.
 
specops said:
Since so many of you are eager to point out: yes, the tested 780Ti is a custom factory overclocked card. Now, get my point: it is not running max clocks but rather the factory standard clocks. It was not overclocked by [H] whereas the Vapor-X was. See? Two factory overclocked cards, only one of them further overclocked to max stable clocks. Not consistent.
Click to expand...

Perhaps if you could read a bit better, So many of us wouldn't have to point on the same thing to you. Let's try it once again. Stay with me here:

From the MSI 780TI review:

For our overclocking efforts, we set the power limit to +5% (the maximum allowed), the voltage to +75mV and spun up the fans to 100% and started to dial up the clock speeds. We began by adjusting the boost clock within Afterburner until we began experiencing instability (note that we had to unlock the voltage modification option within Afterburner and reboot in order to gain access to that functionality). We were able to do some gaming with the base clock set to 1080MHz which resulted in an observed boost clock of 1226MHz, but unfortunately, it was not very stable at that speed. We backed down on the GPU speed until it stabilized at 1055MHz. This resulted in an observed GPU clock of 1176MHz (though it was not uncommon for it to drop back to 1150MHz from time to time). After the GPU overclock was set, we began increasing the memory speed until stability was lost. We were able to achieve an effective 7.4GHz GDDR5 memory rate before the screen began glitching and games begin crashing.

The final overclock of the MSI GeForce GTX 780 Ti GAMING 3G achieved an observed GPU clock increase of 26MHz and a memory increase of 400MHz GDDR5 effective rate.
Click to expand...

Note the results of the MSI card's poor O/C'ing ability above the factory clocks (I even highlighted the important part in bold text)..


So we see from the MSI review that it was pushed 99% to it's limit out of the box on the core, with [H]ard only able to raise the base clock a whopping 26Mhz.

From the Sapphire Vapor-X 290X review:

Note 1 - The top card is the SAPPHIRE Vapor-X R9 290X Tri-X OC video card overclocked to 1180MHz/5.9GHz in all our gameplay testing.

Note 2 - The MSI GTX 780 Ti GAMING video card is running in OC Mode. The actual in-game frequency is a high 1162MHz in all our gameplay tests.

Note 3 - The reference AMD R9 290X is running in Uber Mode.
Click to expand...

So we now see that during the heat to head testing, the MSI 780TI Gaming was in fact O/C'd with the card seeing 1162Mhz, with drops back to 1150Mhz (appears to be thermal throttling at certain points).

Now that I have taken the time to lay all of information you could have gotten from the numerous replies earlier in the thread, or if you took 30 seconds to read the reviews yourself, I hope things are finally clicking. You seem to have this attitude like [H]ard and the those of us here are somehow hating on the 780TI and trying to paint it in an unfavorable light vs the (clearly) superior Vapor-X 290X.
 
In before someone says,

"That's a crappy O/C. I've had 13 of those cards and my worst O/C was 1300MHz and AMD cards throttle. nVidia cards never throttle." :D
 
OMG. That is all I can say. OMG.

So there is now how many of you guys who are not able to get a simple point.

Read carefully:

I would have liked to see the manual max OC numbers of both the 290X and the 780Ti IN THE SAME GRAPHS. You know, as a direct comparison, both running max stable clocks, the latest drivers, no ifs ands or buts. Did this test have those? No, it did not.

Jeez. I guess I'm outta here.
 
you never mentioned that you expected the latest drivers to magically allow the card to overclock beyond what [H] was able to achieve previously.
 
specops said:
OMG. That is all I can say. OMG.

So there is now how many of you guys who are not able to get a simple point.

Read carefully:

I would have liked to see the manual max OC numbers of both the 290X and the 780Ti IN THE SAME GRAPHS. You know, as a direct comparison, both running max stable clocks, the latest drivers, no ifs ands or buts. Did this test have those? No, it did not.

Jeez. I guess I'm outta here.
Click to expand...

Please stay out of the AMD area prime1
 
Can anyone at the H look into (if you want to) about the disconnection between Sapphire's warranty procedures (Which states that warranty starts with the place of purchase) and Amazon's support team.

I had an issue where a card within the 2 year period had an issue, I eventually did get a successful return but I had to prove a few times that Sapphire states the seller is to take the card back for warranty purposes.
 
So this really got ridiculous enough that I'm once more, with minute detail, going to express my feedback.

This was a test of a factory overclocked 290X. It was also overclocked further manually. It was compared to a 780Ti Gaming, which was running its stock clocks. Now some of you guys seem to be confusing the article mentioning "OC mode" for manual overclocking. It is not that at all. The OC mode is the fastest stock setting for the Gaming card, stated by MSI to be 1020/1085MHz (same as [H] stated for base/boost). Those are then boosted further by Kepler boost to a clock rate that varies from card to card. For this individual card, that clock rate (the actual in-game frequency) is 1162MHz. For the 780Ti Gaming card that [H] tested in February, the actual in-game frequency was 1150MHz. Both of those are still stock core frequencies. Then there is the memory clocks which is clearly stated to be stock for the 780Ti in this review. Summary: that is stock clocks for the 780Ti, compared to the manual max OC for the 290X.

[H] has published many manual OC results for GK110 and many manual OC numbers for the Hawaii. That is not the issue. The issue is that there is no direct comparison done between the two with the exception of the 4K CF/SLI article.

The drivers don't help with overclocking but they do help in performance. That is why a fresh set of manual OC numbers would have been nice to have. Also, going by the difference in Kepler boost clocks, the individual 780Ti in this review was not the same as the one [H] tested in February. So there is no telling how much of OC headroom there was left in it.

I asked [H] for some additional info that IMHO would have made the review better. The way you guys jumped on me tells that either you don't know enough about the Kepler GPUs to realize that the 780Ti in this review was not manually overclocked or that you prefer not to have comparative max OC numbers between the two cards. Either way, the joke is on you.
 
Hey Manufacturers

AMD is RED
NVIDIA is GREEN.

Don't put green lights on a AMD product and vice versa.

The card is great except my entire case is red based. This card would throw a wrench in my theme. I know it's a trivial point, but if you are going to dolly it up to look pretty in a windowed case, DO IT RIGHT. You could have replaced the blue, yellow and green leds with red leds of various intensity, or led meters even. (Like a circle of red LEDs around the fan shroud to indicate temperature, RPM, fan speed, power etc... with little hash marks...both functional and appealing)
 
Last edited by a moderator:
specops said:
So this really got ridiculous enough that I'm once more, with minute detail, going to express my feedback.

This was a test of a factory overclocked 290X. It was also overclocked further manually. It was compared to a 780Ti Gaming, which was running its stock clocks. Now some of you guys seem to be confusing the article mentioning "OC mode" for manual overclocking. It is not that at all. The OC mode is the fastest stock setting for the Gaming card, stated by MSI to be 1020/1085MHz (same as [H] stated for base/boost). Those are then boosted further by Kepler boost to a clock rate that varies from card to card. For this individual card, that clock rate (the actual in-game frequency) is 1162MHz. For the 780Ti Gaming card that [H] tested in February, the actual in-game frequency was 1150MHz. Both of those are still stock core frequencies. Then there is the memory clocks which is clearly stated to be stock for the 780Ti in this review. Summary: that is stock clocks for the 780Ti, compared to the manual max OC for the 290X.

[H] has published many manual OC results for GK110 and many manual OC numbers for the Hawaii. That is not the issue. The issue is that there is no direct comparison done between the two with the exception of the 4K CF/SLI article.

The drivers don't help with overclocking but they do help in performance. That is why a fresh set of manual OC numbers would have been nice to have. Also, going by the difference in Kepler boost clocks, the individual 780Ti in this review was not the same as the one [H] tested in February. So there is no telling how much of OC headroom there was left in it.

I asked [H] for some additional info that IMHO would have made the review better. The way you guys jumped on me tells that either you don't know enough about the Kepler GPUs to realize that the 780Ti in this review was not manually overclocked or that you prefer not to have comparative max OC numbers between the two cards. Either way, the joke is on you.
Click to expand...

Just stop whining. You are making yourself look childish by spewing the same stuff over and over again.
 
lucidrenegade said:
Sorry, won't ever buy anything from Sapphire. Their tech support is horrible.
Click to expand...

You have access to [H] and you need tech support why?

FFS man, Google > any tech support for the past 7+ years unless you're lazy as all hell...

Now if you meant their RMA Dept, then I agree and apologize :D
 
Adidas4275 said:
did you test it in crossfire and anything strange with UEFI boot behavior when running 3 monitors? like not showing the bios or black screening?

no black screening when OCing or gaming at all?
Click to expand...

Crossfire can be tricky with these, I currently have 2 vapor x cards and a reference 290x, I intended to go vapor x crossfire but due to the triple fan design and the fact it dumps heat into the case the top card exceeds 80c and becomes very audible (at least for eyefinity resolutions, for something like 1080p they're silent). Suppose it might be workable if you could space them out enough but for me and the motherboard I have its a no-go. 1 of them going back and awaiting corsairs hg10 bracket to liquid cool my reference card.
 
Chris_B said:
Crossfire can be tricky with these, I currently have 2 vapor x cards and a reference 290x, I intended to go vapor x crossfire but due to the triple fan design and the fact it dumps heat into the case the top card exceeds 80c and becomes very audible (at least for eyefinity resolutions, for something like 1080p they're silent). Suppose it might be workable if you could space them out enough but for me and the motherboard I have its a no-go. 1 of them going back and awaiting corsairs hg10 bracket to liquid cool my reference card.
Click to expand...

Yep, you need great airflow throughout your case (and that includes a nice gap between the two cards on the motherboard) to utilise those coolers efficiently and effectively in crossfire.
 
Gumbi10 said:
Yep, you need great airflow throughout your case (and that includes a nice gap between the two cards on the motherboard) to utilise those coolers efficiently and effectively in crossfire.
Click to expand...

I have been using the SilverStone Raven cases for years now for the simple fact that these cases give awesome airflow across video cards.
 
You must log in or register to reply here.
Back
Top