SieB
New member
It's finally here!!!!
Here are some pics of the card itself.
Game benches and overclocking results below.
First impressions though is that it's very quiet
Noticeably more than my 7950 DCUII.
Gaming Benchmarks
Testing Setup and Testing Method
System
Gigabyte Z68 UD3H
I5 2500k
8GB 1600mhz RAM
Sapphire R9 290 Tri-X
Testing Method
For benchmarking games I use the FRAPS MinaMaxAvg benchmark. This benchmarks creates 1 or 2 files depending on whether or not you have MS Excel installed or not.
If you don't have Excel installed FRAPS creates a .txt file which contains the minimum, average and maximum framerate of the game you just played.
If you have Excel installed it also creates a .csv file which can be opened with Excel and used to make a graph. This is the option I used and I made a graph using the results from the FRAPS benchmark.
As for the testing itself I ran each game between 3-5 mins to reflect actual gameplay with various amounts of load on the GPU and used the FRAPS benchmark to capture the min, max and avg FPS.
For each game I have recorded the exact sequence I played through to get the FPS results. It's possible to get different results from different parts of a game so by recording this, anyone who wants to compare results from their cards can run the exact same sequence and settings with there being no inconsistencies apart from the actual hardware they are using.
Just a note as well, the vids were recorded separate from the benchmark runs.
So what I will do is post a vid of the sequence I played to get the results, followed by any extra info and then a graph showing the results
For BF4 I ran Ultra settings with everything maxed out at 100 res scale. I didn't want to increase the res scale to keep it fair when comparing other cards. The reason I chose to use the SP instead of the MP is for the same reasons as above, to keep it consistent and so results can easily be replicated.
A note on this test sequence: If you run this sequence remember to end the benchmark as the ledge you are holding at the end breaks away and you start to fall. Otherwise you get a black screen between scene changes which can cause your FPS to go well above what you would get in game.
For Tomb Raider I used max settings with 4xSSAA, I also tested with both TressFX on and TressFX off.
A note on this test sequence: if you run this sequence remember to end the benchmark before you exit the tunnel at the at the end. Otherwise you get a black screen between scene changes which can cause your FPS to go well above what you would get in game.
For Crysis 3 I tested the highest possible settings possible along with 8xMSAA.
As you can see neither card are capable of a enjoyable gaming experience even though the 290 obviously would give you the better one.
So what I did was also test max settings but with FXAA to represent a good gaming experience, or near enough on both cards.
As with Cryisis above, I tested at the highest possible settings with 4xSSAA.
Same story as before as with Crysis, neither card showed results of a enjoyable experience so I then tested without SSAA to represent a good one.
For Bioshock Infinite I just used the benchmark tool.
Instead of using the results the benchmark gave me though I used the FRAPS benchmark results. For some reason the benchmark results seemed off and were telling me my max FPS was 500+. The FRAPS results, as you can see in the graph look a lot more reasonable.
With The Witcher 2: Assassin of Kings I ran everything maxed out with Ubersampling on then again with Ubersampling off.
Protip: Ubersampling is a GPU killer
Performance Benchmarks
Nothing special about how or what I did for these, I simply ran them and took a screenshot of the results.
For benchmarks I used Unigine's Valley and Heaven, 3Dmark's Ice Strike Extreme, Cloud Gate and Fire Storm Extreme and finally Catzilla.
My CPU is at stock as well as my GPU because I wanted this to be based on the performance from the stock GPU without the CPU adding to it.
So on with the overclocking, before I start I just want to mention that I am just your average user just like many of you and when it comes to overclocking my knowledge is very basic.
For this reason I have not gone all out to break records or achieve the highest possible overclock and have stuck within the safe limits of the card and what the average user is capable of achieving.
So first up we have performance benchmarks, the ones I used are the same as the ones I used previously, minus 3DMark's Ice Storm and Cloud Gate. So the ones I did use are 3DMark Fire Strike Extreme, Unigine's Valley and Heaven, and last but not least Catzilla.
For overclocking my GPU I used Sapphire Trixx, me and MSI Afterburner have never got on so I stuck with Trixx because it's what I have had the best results with.
After playing around with clocks and using 3DMark's Fire Strike Extreme to find my maximum stable OC, I managed to achieve 1200mhz on the core and 1550mhz on the mem. However, performance in 3DMark began to drop off at these clocks so I decided to back off and lower them to 1190mhz on the core and 1500mhz on the mem at 1.27v. This is the OC I decided to stick with and is what I used for my max Benchmarking OC.
Going from stock clocks to my max bench OC netted me an extra 465 points in Fire Strike Extreme and pull in front of what I assume (could be OC'd but doesn't say) is a stock GTX Titan by 179 points
With Unigine's Heaven I was able to improve my score from 1286 to 1451 giving me an increase of 165 points.
As with Heaven, I was also able to increase my score in Valley by quite a bit from 2390 to 2573 giving me an increase of 183 points.
So next up is Catzilla which saw the most increase in score from my OC going from 10162 to 11558 giving me an increase of 1396 points.
It also scored me 3rd place on the leaderboards for 1080p with a single R9 290
http://www.catzilla.com/toplist?pag...lti=s&ven=a&ctype=d&manual=r9+290&submit=Show
Gaming OC Benchmarks
For gaming I decided to lower the clocks to a more 24/7 average user safe OC.
The clocks I went with are 1150mhz on the core and because of the lower core clock I decided to up the mem a bit and went with 1550mhz on the memory.
For the results I ran the same sequence as I did in my stock results posted above using the same method to record the min, max and average FPS.
Not much of a performance increase in Crysis 3 although the few FPS extra it did give made the game run a fair bit smother and more enjoyable at max settings compared to stock.
As with Crysis 3 there were no major performance gains in Metro Last Light. With Crysis 3 though the increase did improve the palyability at max settings but with Metro there was no noticeable change.
Bioshock Infinite was the diamond in the rough when it came to an increase in FPS with quite impressive gains.
So that concludes the overclocking, as mentioned at the start of this section i'm not much of an overclocker and my knowledge is basic. Higher overclocks with this card are definitely possible, especially on water.
There were no real gains to be had in the games I tested but I can not speak for all games. Some games may benefit from the overclock I went with.
Benchmarks on the other hand saw a good improvement from stock and with a bit of tweaking and pushing the volts some more, higher scores are guaranteed on my card and most likely on others of the same kind.
Personally, I don't see the point in overclocking a GPU for games as the performance increase is minimal and the increase in temps and noise isn't worth it.
Also by the time I need to overclock a GPU in order to be able to play a game then it's time to upgrade.
As for benching, it depends on how serious of a overclocker you are. If you are someone that runs the odd benchmark here and there then having a good overclocking card isn't going to be a big deal.
The only type of person GPU overclocking really applies to are people who are constantly running benchmarks and trying to achieve the highest score they can.
That is just my opinion on GPU overclocking, others may have a different one but each to their own.
So we all know that the only downfall of the 290 reference cards was the terrible stock AMD cooler.
Not only was it loud and and did a poor job of cooling the cards, it also caused them to throttle due to thermal limits resulting in the core clock to get lowered and performance and overclocking was gimped.
So how does the Sapphire Tri-X cooler do? Does it solve all of the above problems?
Well, as you can see from my performance results above, performance is fixed. The core clock remains stable and no thermal throttling is taking place, a good way of showing this is if you look at my Catzilla benchmark results. The green line on the graph represents the core clock, and as you can see on both my stock test and overclock test the core clock remains strait throughout with no dips apart from where there were scene changes and load screens.
So what about the temps?
What I did to test temps was to test under two different scenarios. One that represents the max gaming temp and one that represents the max benching temp.
For the gaming scenario I used one of the most demanding games that stresses the GPU to the max, Crysis 3.
And for the benching scenario I used one of the most card melting and heat producing benchmars A.K.A Furmark.
Before starting these test I took a reading of my ambient temp which was a toasty 21*c due to the heating being on full whack in my flat.
To measure the temps while gaming I used Cupid Hardware Monitor which gives a reading for the min and max temps of the GPU.
I then played Crysis 3 for 30mins and here is the result of my min and max temps.
For Furmark I used the built in 15min benchmark option. during this the max temp of the card reached 80*c.
Now this under Furmark and in no way under a normal gaming situation would the card get this hot. This is just to show how when pushed as hard as possible how hot the card gets and how well the cooler cools it.
Noise
For this part you will have to take my word for it as I have no way of showing you how loud it is without you actually being here.
The card, even when gaming in Crysis 3 was inaudible, thats not to say it wasn't making any noise just that I didn't notice it making any.
With my 7950, as soon as it started to get hot I could hear the fans kick in. Even then my 7950 wasn't loud, you could tell it was running but it wasn't loud.
With my 290 I am yet to notice it running whilst gaming, it's not like where I could tell when my 7950 was getting hot. With the 290 I can't hear it at all compared to how loud it is when running idle.
During benchmarks and overclocking though it is a different story, as the card reaches ~80*c the fans really start to ramp up. Although you could lower this noise quite a bit by making a custom fan profile so the card doesn't get so hot.
Benching and overclocking don't represent real gaming though and under a normal gaming situation the card is very quiet.
With the GPU market in the state it's in at the moment, there is very little in performance between cards with the only real difference being price.
The 290X being £100 more than a non X 290 with only around 2-3 FPS difference and a GTX 780Ti being up to £250 more than a non ref 780 or 290 with only 10% the performance, the 290 is a great bang for your buck card. Don't get me wrong, the 780Ti is an excellent card and a great performer, it's just ridiculously priced in my opinion especially compared to the competition.
Now, with all problems of the original stock AMD 290 with a ref AMD cooler solved, there really isn't any reason not to get one.
Saying that though, the 780 is the same price and is just as fast. For me it was a difficult choice between the 780 and 290, but in the end it came down to me wanting to see how Mantle plays out so I went with the 290.
And I am glad I did go with the 290 now TBH, the Sapphire Tri-X is a excellent cooler and the card itself is a very good performer with a decent amount of overclocking headroom. At only £380 it competes almost equally or better than a 290X or 780, but it also competes well with the 780Ti with it's great price to performance.
Thank you to all that have read this, it has been fun doing it. Hopefully it gives you the information you may need if you are considering buying a 290 yourself.
Here are some pics of the card itself.
Game benches and overclocking results below.
First impressions though is that it's very quiet
Noticeably more than my 7950 DCUII.
Gaming Benchmarks
Testing Setup and Testing Method
System
Gigabyte Z68 UD3H
I5 2500k
8GB 1600mhz RAM
Sapphire R9 290 Tri-X
Testing Method
For benchmarking games I use the FRAPS MinaMaxAvg benchmark. This benchmarks creates 1 or 2 files depending on whether or not you have MS Excel installed or not.
If you don't have Excel installed FRAPS creates a .txt file which contains the minimum, average and maximum framerate of the game you just played.
If you have Excel installed it also creates a .csv file which can be opened with Excel and used to make a graph. This is the option I used and I made a graph using the results from the FRAPS benchmark.
As for the testing itself I ran each game between 3-5 mins to reflect actual gameplay with various amounts of load on the GPU and used the FRAPS benchmark to capture the min, max and avg FPS.
For each game I have recorded the exact sequence I played through to get the FPS results. It's possible to get different results from different parts of a game so by recording this, anyone who wants to compare results from their cards can run the exact same sequence and settings with there being no inconsistencies apart from the actual hardware they are using.
Just a note as well, the vids were recorded separate from the benchmark runs.
So what I will do is post a vid of the sequence I played to get the results, followed by any extra info and then a graph showing the results
For BF4 I ran Ultra settings with everything maxed out at 100 res scale. I didn't want to increase the res scale to keep it fair when comparing other cards. The reason I chose to use the SP instead of the MP is for the same reasons as above, to keep it consistent and so results can easily be replicated.
A note on this test sequence: If you run this sequence remember to end the benchmark as the ledge you are holding at the end breaks away and you start to fall. Otherwise you get a black screen between scene changes which can cause your FPS to go well above what you would get in game.
For Tomb Raider I used max settings with 4xSSAA, I also tested with both TressFX on and TressFX off.
A note on this test sequence: if you run this sequence remember to end the benchmark before you exit the tunnel at the at the end. Otherwise you get a black screen between scene changes which can cause your FPS to go well above what you would get in game.
For Crysis 3 I tested the highest possible settings possible along with 8xMSAA.
As you can see neither card are capable of a enjoyable gaming experience even though the 290 obviously would give you the better one.
So what I did was also test max settings but with FXAA to represent a good gaming experience, or near enough on both cards.
As with Cryisis above, I tested at the highest possible settings with 4xSSAA.
Same story as before as with Crysis, neither card showed results of a enjoyable experience so I then tested without SSAA to represent a good one.
For Bioshock Infinite I just used the benchmark tool.
Instead of using the results the benchmark gave me though I used the FRAPS benchmark results. For some reason the benchmark results seemed off and were telling me my max FPS was 500+. The FRAPS results, as you can see in the graph look a lot more reasonable.
With The Witcher 2: Assassin of Kings I ran everything maxed out with Ubersampling on then again with Ubersampling off.
Protip: Ubersampling is a GPU killer
Performance Benchmarks
Nothing special about how or what I did for these, I simply ran them and took a screenshot of the results.
For benchmarks I used Unigine's Valley and Heaven, 3Dmark's Ice Strike Extreme, Cloud Gate and Fire Storm Extreme and finally Catzilla.
My CPU is at stock as well as my GPU because I wanted this to be based on the performance from the stock GPU without the CPU adding to it.
Overclocking, Noise and Temps
So on with the overclocking, before I start I just want to mention that I am just your average user just like many of you and when it comes to overclocking my knowledge is very basic.
For this reason I have not gone all out to break records or achieve the highest possible overclock and have stuck within the safe limits of the card and what the average user is capable of achieving.
Performance Benchmarks
So first up we have performance benchmarks, the ones I used are the same as the ones I used previously, minus 3DMark's Ice Storm and Cloud Gate. So the ones I did use are 3DMark Fire Strike Extreme, Unigine's Valley and Heaven, and last but not least Catzilla.
For overclocking my GPU I used Sapphire Trixx, me and MSI Afterburner have never got on so I stuck with Trixx because it's what I have had the best results with.
After playing around with clocks and using 3DMark's Fire Strike Extreme to find my maximum stable OC, I managed to achieve 1200mhz on the core and 1550mhz on the mem. However, performance in 3DMark began to drop off at these clocks so I decided to back off and lower them to 1190mhz on the core and 1500mhz on the mem at 1.27v. This is the OC I decided to stick with and is what I used for my max Benchmarking OC.
Going from stock clocks to my max bench OC netted me an extra 465 points in Fire Strike Extreme and pull in front of what I assume (could be OC'd but doesn't say) is a stock GTX Titan by 179 points
With Unigine's Heaven I was able to improve my score from 1286 to 1451 giving me an increase of 165 points.
As with Heaven, I was also able to increase my score in Valley by quite a bit from 2390 to 2573 giving me an increase of 183 points.
So next up is Catzilla which saw the most increase in score from my OC going from 10162 to 11558 giving me an increase of 1396 points.
It also scored me 3rd place on the leaderboards for 1080p with a single R9 290
http://www.catzilla.com/toplist?pag...lti=s&ven=a&ctype=d&manual=r9+290&submit=Show
Gaming OC Benchmarks
For gaming I decided to lower the clocks to a more 24/7 average user safe OC.
The clocks I went with are 1150mhz on the core and because of the lower core clock I decided to up the mem a bit and went with 1550mhz on the memory.
For the results I ran the same sequence as I did in my stock results posted above using the same method to record the min, max and average FPS.
Not much of a performance increase in Crysis 3 although the few FPS extra it did give made the game run a fair bit smother and more enjoyable at max settings compared to stock.
As with Crysis 3 there were no major performance gains in Metro Last Light. With Crysis 3 though the increase did improve the palyability at max settings but with Metro there was no noticeable change.
Bioshock Infinite was the diamond in the rough when it came to an increase in FPS with quite impressive gains.
So that concludes the overclocking, as mentioned at the start of this section i'm not much of an overclocker and my knowledge is basic. Higher overclocks with this card are definitely possible, especially on water.
There were no real gains to be had in the games I tested but I can not speak for all games. Some games may benefit from the overclock I went with.
Benchmarks on the other hand saw a good improvement from stock and with a bit of tweaking and pushing the volts some more, higher scores are guaranteed on my card and most likely on others of the same kind.
Personally, I don't see the point in overclocking a GPU for games as the performance increase is minimal and the increase in temps and noise isn't worth it.
Also by the time I need to overclock a GPU in order to be able to play a game then it's time to upgrade.
As for benching, it depends on how serious of a overclocker you are. If you are someone that runs the odd benchmark here and there then having a good overclocking card isn't going to be a big deal.
The only type of person GPU overclocking really applies to are people who are constantly running benchmarks and trying to achieve the highest score they can.
That is just my opinion on GPU overclocking, others may have a different one but each to their own.
Noise and Temps
Tempratures
So we all know that the only downfall of the 290 reference cards was the terrible stock AMD cooler.
Not only was it loud and and did a poor job of cooling the cards, it also caused them to throttle due to thermal limits resulting in the core clock to get lowered and performance and overclocking was gimped.
So how does the Sapphire Tri-X cooler do? Does it solve all of the above problems?
Well, as you can see from my performance results above, performance is fixed. The core clock remains stable and no thermal throttling is taking place, a good way of showing this is if you look at my Catzilla benchmark results. The green line on the graph represents the core clock, and as you can see on both my stock test and overclock test the core clock remains strait throughout with no dips apart from where there were scene changes and load screens.
So what about the temps?
What I did to test temps was to test under two different scenarios. One that represents the max gaming temp and one that represents the max benching temp.
For the gaming scenario I used one of the most demanding games that stresses the GPU to the max, Crysis 3.
And for the benching scenario I used one of the most card melting and heat producing benchmars A.K.A Furmark.
Before starting these test I took a reading of my ambient temp which was a toasty 21*c due to the heating being on full whack in my flat.
To measure the temps while gaming I used Cupid Hardware Monitor which gives a reading for the min and max temps of the GPU.
I then played Crysis 3 for 30mins and here is the result of my min and max temps.
For Furmark I used the built in 15min benchmark option. during this the max temp of the card reached 80*c.
Now this under Furmark and in no way under a normal gaming situation would the card get this hot. This is just to show how when pushed as hard as possible how hot the card gets and how well the cooler cools it.
Noise
For this part you will have to take my word for it as I have no way of showing you how loud it is without you actually being here.
The card, even when gaming in Crysis 3 was inaudible, thats not to say it wasn't making any noise just that I didn't notice it making any.
With my 7950, as soon as it started to get hot I could hear the fans kick in. Even then my 7950 wasn't loud, you could tell it was running but it wasn't loud.
With my 290 I am yet to notice it running whilst gaming, it's not like where I could tell when my 7950 was getting hot. With the 290 I can't hear it at all compared to how loud it is when running idle.
During benchmarks and overclocking though it is a different story, as the card reaches ~80*c the fans really start to ramp up. Although you could lower this noise quite a bit by making a custom fan profile so the card doesn't get so hot.
Benching and overclocking don't represent real gaming though and under a normal gaming situation the card is very quiet.
Conclusion
With the GPU market in the state it's in at the moment, there is very little in performance between cards with the only real difference being price.
The 290X being £100 more than a non X 290 with only around 2-3 FPS difference and a GTX 780Ti being up to £250 more than a non ref 780 or 290 with only 10% the performance, the 290 is a great bang for your buck card. Don't get me wrong, the 780Ti is an excellent card and a great performer, it's just ridiculously priced in my opinion especially compared to the competition.
Now, with all problems of the original stock AMD 290 with a ref AMD cooler solved, there really isn't any reason not to get one.
Saying that though, the 780 is the same price and is just as fast. For me it was a difficult choice between the 780 and 290, but in the end it came down to me wanting to see how Mantle plays out so I went with the 290.
And I am glad I did go with the 290 now TBH, the Sapphire Tri-X is a excellent cooler and the card itself is a very good performer with a decent amount of overclocking headroom. At only £380 it competes almost equally or better than a 290X or 780, but it also competes well with the 780Ti with it's great price to performance.
Thank you to all that have read this, it has been fun doing it. Hopefully it gives you the information you may need if you are considering buying a 290 yourself.
Last edited:
