If I had the GPUs, I'd test them, but I don't have any 780 Ti or Titan, just like I don't have any of the 750-765 GPUs, R9 290, or most of the R7 options. As TViceman notes, 780 Ti should be pretty close to the 980 in performance -- it might not even be 8% slower, depending on the game. 780 and 970 basically overlap enough that testing both is a bit redundant, and long-term I might drop the 780, but it's interesting to see how the two track over a variety of games.
Jarred I appreciate all the work you do. Testing each card at several resolutions and quality settings takes hours and can definitely be a tedious process.
One more thing! Many sites report stability issues because of the DRM protection. Did you have any of that? I am about to decide to wait for new patch to solve these problems or just go and jump in to buy.
Yeah? Well then how about doing just one gpu and then subtract an arbitrary percentage based on a guesstimate to figure out all the others? Heck, why even do a single one and not just take a guess based on the game engine and save yourself the trouble of benchmarking entirely? Also, watch your mouth - there really is no need for petty name-calling. It only makes you appear as an immature individual.
Just what we need more "herd mechanics" and censorship, by the overtly emotional crowd that clicks and reposts 3 times when so personally offended and enraged, and actually believes their " clicky down vote" is a valuable "opposing opinion" that someone else for some unknown reason would be interested in, not to mention the exorbitant laziness and lack of succinct or even discernible counter point the down vote contains.
So, the owners are supposed to provide a selfish outlet for your childish emotions, so you can move on from a post you disagree with, satisfied as easily and expediently as possible without any benefit for the fellow readers ?
You should write here your system specs, hardware used and frequencies for memory, cpu and gpu. Also inform Anandtech about your windows version, settings, optimizations and other software you are running in the background. That will help them create a clone of your system that will produce numbers very close to yours. I can understand perfectly your problem. Trying to estimate based on these charts, your cards performance with an accuracy of 0.1 fps using kindergarten arithmetic, it must be a very difficult task.
The 780ti was the top-of-the-line for its generation. Bypassing it entirely and instead going with #2 and #3 of that generation isn't how AT usually do their thing. Nonetheless, Jarred was kind enough to explain the reason behind the discrepancy, so you really didn't have to take a stand in his defence.
I was sarcastic indeed. Everyone wants to see his card or cpu in the charts. Well it is just not possible. Should I add here the cpu+gpu combinations. Maybe also the various speeds that 780ti can archive with overclocking on the gpu and/or memory?
Looking at 780 and 980 you can have a pretty good idea where 780ti falls. It's not rocket science. In fact you don't even need kindergarten arithmetic to have an idea how 780ti performs.
So yes, I was sarcastic. You know what I think? I think he was hopping to see a couple of charts where 780Ti actually beats 980. That's why he wanted that model included.
"You know what I think? I think he was hopping to see a couple of charts where 780Ti actually beats 980. That's why he wanted that model included."
lol, not really. As I explained earlier, the Ti variant being the top of the crop in its generation, I expected its inclusion in the bench, nothing more... the 780ti isn't the first GPU I've bought, and it wouldn't be the last, so if anything, I would've been a bit disappointed if it beat the 980.
Only when AMD has a high holy crown is it important for it to be included. Other much more important things than video cards to test are Basement Home Theatres and things like that for the elite.
Do you have any idea how long running multiple tests (to average out) for each combintion of the four resolutions and twelve GPUs and the author has chosen would take? Because it is a long time man, like a shocking 'in no way can this be worth ajournalist pay' amont of time.
Also: nice work showing both min and avg FPS, wish everyone did that.
I think in the latest Graphics Card Buyer guide you said 2gb of VRAM would be more then enough. I am not sure anymore , I hope Ubisoft AC Unity is a exception but for all we know games aimed for XBox One and PS4 might take the easy route when it comes to allocating memory after beeing ported to PC?Because AC: Unity recommended system requirements a GTX780 3GB VRAM and Miminmum GTX 680 2gb VRAM bonkers me compared to Lors of the Fallen fair system requirements.
Time will tell ,... curious how this generation of optimizing for console will reflect on PC.
AC Unity is unoptimized as hell. I have tried it. No matter what resolution you play at. Struggles at the same scenes. I patch would be a must go now out i would say.
I think turning down the texture quality is one option for PCs with less VRAM, but it's not going to be ideal in all cases. With the consoles now having 8GB, though, 4GB is probably the safer bet on PC GPUs.
Yeah was thinkign the same that if you can ,you should aim for 4gb just to be safe. You can get most midrange cards with a 4gb option but default they are still 2GB or sometimes 3GB. Something to note maby in the next video card buyers guide :)
"Need I mention that this is an NVIDIA "The Way It's Meant To Be Played" title?"
Nope! As usual I gone to see the charts first, read the text later. From the first chart I gone back to the text and started looking for the word "Nvidia" in the article.
Funny how AMD sponsored games work fine with any hardware while Nvidia sponsored games always have major stability or performance problems with AMD hardware.
I think the bigger issue is more performance and stability on AMD in general. NVIDIA has more driver engineers than AMD and they tend to work with more gaming companies; even for AMD titles (e.g. Civ:BE), there was a "Game Ready" driver available before the game shipped. AMD did better a few years back, but it seems they're falling off the pace a bit right now.
Any chance to the with different cpus? Take a look on AC Unity : http://postimg.org/image/meg9uhaxp/ AMD cpu is no go. Many games could be affacted as well. Short note: i do have AMD cpu, not one but 5.
Unfortunately I don't have any AMD mobo/APU to do the testing on right now, but it wouldn't surprise me to see CPU bottlenecks come into play on lower performance CPUs/APUs. The CPU bottleneck appears to be around 110-115 FPS on my 4.1GHz i7-4770K (at Low/Medium quality), so at higher quality and with dual GPUs it would probably be lower than that. I think most people would be fine with a single fast GPU and an APU, but breaking 60 FPS might require something more potent.
On one hand AMD is letting people go, so maybe less programmers in the drivers team. On the other hand many games are starting to be made with AMD hardware in mind thanks to the consoles. This two things go to opposite directions. If AMD in a year or two, looks better in the driver area, then it wasn't the number of engineers as big a factor as we both think, but the way games where made. Games where made on Intel+Nvidia hardware from what I remember and just tested for compatibility on AMD hardware. So Nvidia had to address less problems than AMD. Of course in the above though Ubisoft is the exception considering their tight relationship with Nvidia. They will always program games with Nvidia hardware in mind. But what I want to say is that. In a year or two, if AMD with less stuff in the programmers team is starting to look better, then those conspiracy theories I am talking about now, could have a very large dose of reality.
AMD has been letting people go for years. AMD already had the lesser of the people, because any one with a brain wanted to work for the company with BILLIONS in the bank, nVidia.
All the conspiracy theories in the world pale in comparison to the common sense dollar analysis, call it the terrible social consequences of capitalism for amd fans...
Not too mention, AMD is to be pitied, it's end user fan base is constantly demanding more for less and is the worst of the penny pinching freak jobs, the exact kind of customer and consumer every decent businessman wants to avoid 24/7/365... as they drain the viability of the company.
Scrimping little penniless tightwad complainers... always demanding a better deal - the only upside being the insane loyalty and "for the poor underdog" mindset that can be used and wielded as a powerful yet dishonest PR ax. Well, in the end, that spells destruction, and in the mean time, it spells under performance, problems, lagging, and inferior quality in every area.
Maybe the FED can print up a multi billion dollar bailout "for free".
That isn't even true, take Dragon Age and it's titles for instance. AMD is the one pulling stunts to strip the competition of performance when one of their 4 "game devs" sends off email answers to game companies.
Nvidia has literally 100 traversing the world, AMD had 4 last time I saw it publicly revealed. So there's only so much AMD can do with 4 underpaid disgruntled and overworked stringers facing the firing ax every time the quarterly loss sheet hits the news.
Maybe the EU should make a new progressive law whereby nVidia must begin optimizing AMD GPU's as "each according to it's need" would please amd fans internationale.
"but I don't think AMD has spent as much time optimizing their drivers for the game"
I'm tired of seeing a need for special application-specific "optimisations" (read: corner-cutting) in the low level drivers just to make the game work. If the games used the APIs correctly, and the drivers implemented the APIs correctly, driver optimisations should benefit all games.
Maybe it's just me, but it's hard to keep up with all of the new GPU naming schemes. For those of us who are a little behind the times, it might be helpful to reference some of the older "equivalents" in articles like this. It would let us more quickly figure out where our existing hardware is likely to fall on these charts.
Doing a bit of searching, I can see that the R7 250X is essentially equivalent to a Radeon HD 7770 (and hence not far off of the older 6770 and 5770). That's useful knowledge! It says that yesterday's "mid-range" videocard is going to have a tough time handling the current generation of games, even at compromised settings.
True. The R9 280 is the same as the HD 7950, and the R9 280X is basically the same as HD 7970 (GHz edition or overclocked a bit). R9 285 is a different core, but performance should be about the same as the R9 280 as well. On the NVIDIA side, there haven't been as many straight rebadges (at least for the GPUs I have), though in terms of performance I'd estimate the 860M as being relatively close to the old GTX 580 in performance. But I could be off there, so don't quote me. ;-)
rebranding is no longer a topic since AMD dove into the lake of fire and lies head first in that area...
I find that extremely unfortunate as all it would take is the old name of the card next to the new name of the card in the charts 7850 7870 9750 ... everyone would realize it's been like years since AMD made anything new - except for 290x hawaii which I'd assume is just a bloated out pitcairn or tahiti.
I have to admit it's amazing their cards are even on the charts - I think AMD has pulled like 30% performance out of the driver pile across the board over time - the image quality has suffered but that is never a topic anymore either because we have to save the drowning underdog, it's a worldwide group effort.
Nice article, waiting for AC:U as it might be my next purchase if it runs OK on PC. Could you maybe also run a few tests with either different CPUs or downclocking / disabling cores on your i7 in 1080p to see if some games may be CPU limited ?
I might be willing to test a couple specific configurations once the main testing is complete, but doing a second CPU with every GPU configuration would double my testing time. Oh, and AC:U is already a pain in the butt to benchmark; there are four times of day that you might be in, and each has slightly different performance characteristics, plus the crowds are rather non-deterministic.
I started testing and then after a few hours I started realizing some of my results didn't make much sense, so I had to go back to the drawing board. Now I'm having to run each test sequence a couple extra times to verify results, but I've finally got reasonably consistent numbers. There's more variance than in other games (I'd say as much as 5-10% between runs on the same hardware), but the I run each setting three times and verify I get at least two scores that are within a couple percent, then I take the highest score out of those.
More than you wanted to know, I'm sure. I'd guess AC:U is a bit less CPU limited than some games (or at least it's looking very GPU limited so far). What GPU do you have? I can at least give you a guess as to how it will run AC:U, though so far I've been testing NVIDIA GPUs.
Jared I would like you to test Dragon Age Inquisition when the game is out . The game looks amazing with the Frostbite 3 engine and its a well reviewed game which could find a place in your testing suite. Call of duty Advanced Warfare is worth testing as the game is being well reviewed and is a popular game. Surprisingly COD AW is being praised for its improved graphics. Finally Farcry 4 is the last game from this holiday season which is worth testing as it will surely be a great looking game and a worthy GPU benchmark title.
btw how is the R9 290X doing so poorly in your review when other reviews show it performing on par with GTX 970 and GTX 780 OC and GTX Titan. Have you picked a corner case in this game where AMD cards do poorly ?
Obviously I don't know what areas other places are using for testing, but I did choose a place where I noticed more of a slowdown -- basically, I look for "worst case" areas for benchmarking, as people usually want to know how things will fare in the most demanding areas. Running round indoors, LotF has much better performance, while outdoors there are some vistas that you encounter where frame rates can take a dive.
If they're getting minimum FPS relatively close to the average, obviously they're not hitting some of the more demanding areas. I didn't bother to try any translation services, so I'm not sure what exactly they're saying or how they're testing, so that's about all I can say.
One thing somewhat interesting to note is that the patch that improved performance on GPUs with 2GB and less RAM actually tended to drop performance a bit on the R9 290X. At 2560x1440 I measured about a 5FPS drop on 290X, while only a ~2FPS drop on 980/970. 1920x1080 Ultra showed a similar drop on 290X. I have no idea why that happened, but it's very much reproducible.
Jared Sorry missed out Middle Earth Shadow of Mordor which is one of the best games of 2014 and is graphically quite demanding. It would be a game worth benchmarking.
I do have some benchmark numbers from the game, but there's a lot of weird stuff with the Shadow of Mordor benchmark. For one, it includes time from before the benchmark starts, so that throws things off. The minimum FPS is also pretty much useless, so you have to use FRAPS to get a better idea of a "true" minimum. At this point it's older so I have other games to test, but maybe a short article would be worth posting if I get a chance.
The comparison of the various quality settings including the visual impact of the graphics and the performance impact was very nice. I would love to see more along these lines.
I have been reading these and appreciate them. I have two R9 290's and I can tell where they'd likely fall if I were to pick up this game. My CPU will cause me to lose some FPS and I'm not crossfiring the cards as my current setup causes them to run really hot in the case (non-blower model).
I have found that I'd like to see double the spacing of PCIe x16 slots for better thermals in a 2 way SLI/Crossfire setup from mobo makers. One card plays fine and the other goes to $#!+ on temps to me - ~10+ Celsius higher when pegged out.
Ok. 1080P ultra 280x 48 290x 63, soooooo 290 is like 55FPS.
Yeah, you should be able to tell, but it's getting worse for accuracy with the cpu offloading, and you didn't say what your cpu is... but just see any 290 bench game w similar cpu reliance , and you have - reduce percentage accordingly.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
53 Comments
Back to Article
D. Lister - Wednesday, November 12, 2014 - link
Thanks for being so "thorough", as an owner of a 780Ti, this article was very "helpful" for me.tviceman - Wednesday, November 12, 2014 - link
Or you could not be a prick. Let me "help" you. Take any given GTX 980 score, subtract 8%, and you get your 780 TI performance.JarredWalton - Wednesday, November 12, 2014 - link
If I had the GPUs, I'd test them, but I don't have any 780 Ti or Titan, just like I don't have any of the 750-765 GPUs, R9 290, or most of the R7 options. As TViceman notes, 780 Ti should be pretty close to the 980 in performance -- it might not even be 8% slower, depending on the game. 780 and 970 basically overlap enough that testing both is a bit redundant, and long-term I might drop the 780, but it's interesting to see how the two track over a variety of games.tviceman - Wednesday, November 12, 2014 - link
Jarred I appreciate all the work you do. Testing each card at several resolutions and quality settings takes hours and can definitely be a tedious process.siriq - Wednesday, November 12, 2014 - link
Nice bench, if i got time i will do on my rig as i did before with civ beyond earth as well. (GTX570)siriq - Wednesday, November 12, 2014 - link
One more thing! Many sites report stability issues because of the DRM protection. Did you have any of that? I am about to decide to wait for new patch to solve these problems or just go and jump in to buy.JarredWalton - Wednesday, November 12, 2014 - link
I haven't had any stability, but I'm using the Steam version. Not sure if that matters?D. Lister - Wednesday, November 12, 2014 - link
"I don't have any 780 Ti"Oh well :). AT should provide its reviewers with all the necessary hardware samples. If nothing else, it adds to their very useful benchmark database.
Death666Angel - Thursday, November 13, 2014 - link
This isn't a review. It's a pipeline article. Adjust your expectations.D. Lister - Wednesday, November 12, 2014 - link
Yeah? Well then how about doing just one gpu and then subtract an arbitrary percentage based on a guesstimate to figure out all the others? Heck, why even do a single one and not just take a guess based on the game engine and save yourself the trouble of benchmarking entirely? Also, watch your mouth - there really is no need for petty name-calling. It only makes you appear as an immature individual.Hairs_ - Thursday, November 13, 2014 - link
Anandtech could really do with a downvote system for comments like this.FlushedBubblyJock - Thursday, November 20, 2014 - link
Just what we need more "herd mechanics" and censorship, by the overtly emotional crowd that clicks and reposts 3 times when so personally offended and enraged, and actually believes their " clicky down vote" is a valuable "opposing opinion" that someone else for some unknown reason would be interested in, not to mention the exorbitant laziness and lack of succinct or even discernible counter point the down vote contains.So, the owners are supposed to provide a selfish outlet for your childish emotions, so you can move on from a post you disagree with, satisfied as easily and expediently as possible without any benefit for the fellow readers ?
Seems that pegs it.
Hairs_ - Thursday, November 13, 2014 - link
Anandtech could really do with a downvote system for comments like this.Hairs_ - Thursday, November 13, 2014 - link
Anandtech could really do with a downvote system for comments like this.D. Lister - Thursday, November 13, 2014 - link
I could swear there was an echo in here.yannigr2 - Wednesday, November 12, 2014 - link
You should write here your system specs, hardware used and frequencies for memory, cpu and gpu. Also inform Anandtech about your windows version, settings, optimizations and other software you are running in the background. That will help them create a clone of your system that will produce numbers very close to yours. I can understand perfectly your problem. Trying to estimate based on these charts, your cards performance with an accuracy of 0.1 fps using kindergarten arithmetic, it must be a very difficult task.D. Lister - Wednesday, November 12, 2014 - link
The 780ti was the top-of-the-line for its generation. Bypassing it entirely and instead going with #2 and #3 of that generation isn't how AT usually do their thing. Nonetheless, Jarred was kind enough to explain the reason behind the discrepancy, so you really didn't have to take a stand in his defence.JarredWalton - Thursday, November 13, 2014 - link
I believe Yannigr2 was being sarcastic towards the original poster, not serious. :-)yannigr2 - Thursday, November 13, 2014 - link
I was sarcastic indeed. Everyone wants to see his card or cpu in the charts. Well it is just not possible. Should I add here the cpu+gpu combinations. Maybe also the various speeds that 780ti can archive with overclocking on the gpu and/or memory?Looking at 780 and 980 you can have a pretty good idea where 780ti falls. It's not rocket science. In fact you don't even need kindergarten arithmetic to have an idea how 780ti performs.
So yes, I was sarcastic. You know what I think? I think he was hopping to see a couple of charts where 780Ti actually beats 980. That's why he wanted that model included.
D. Lister - Thursday, November 13, 2014 - link
@yannigr2"You know what I think? I think he was hopping to see a couple of charts where 780Ti actually beats 980. That's why he wanted that model included."
lol, not really. As I explained earlier, the Ti variant being the top of the crop in its generation, I expected its inclusion in the bench, nothing more... the 780ti isn't the first GPU I've bought, and it wouldn't be the last, so if anything, I would've been a bit disappointed if it beat the 980.
FlushedBubblyJock - Thursday, November 20, 2014 - link
Only when AMD has a high holy crown is it important for it to be included.Other much more important things than video cards to test are Basement Home Theatres and things like that for the elite.
philosofa - Friday, November 14, 2014 - link
Do you have any idea how long running multiple tests (to average out) for each combintion of the four resolutions and twelve GPUs and the author has chosen would take? Because it is a long time man, like a shocking 'in no way can this be worth ajournalist pay' amont of time.Also: nice work showing both min and avg FPS, wish everyone did that.
plopke - Wednesday, November 12, 2014 - link
I think in the latest Graphics Card Buyer guide you said 2gb of VRAM would be more then enough. I am not sure anymore , I hope Ubisoft AC Unity is a exception but for all we know games aimed for XBox One and PS4 might take the easy route when it comes to allocating memory after beeing ported to PC?Because AC: Unity recommended system requirements a GTX780 3GB VRAM and Miminmum GTX 680 2gb VRAM bonkers me compared to Lors of the Fallen fair system requirements.Time will tell ,... curious how this generation of optimizing for console will reflect on PC.
siriq - Wednesday, November 12, 2014 - link
AC Unity is unoptimized as hell. I have tried it. No matter what resolution you play at. Struggles at the same scenes. I patch would be a must go now out i would say.JarredWalton - Wednesday, November 12, 2014 - link
I think turning down the texture quality is one option for PCs with less VRAM, but it's not going to be ideal in all cases. With the consoles now having 8GB, though, 4GB is probably the safer bet on PC GPUs.plopke - Wednesday, November 12, 2014 - link
Yeah was thinkign the same that if you can ,you should aim for 4gb just to be safe. You can get most midrange cards with a 4gb option but default they are still 2GB or sometimes 3GB. Something to note maby in the next video card buyers guide :)yannigr2 - Wednesday, November 12, 2014 - link
"Need I mention that this is an NVIDIA "The Way It's Meant To Be Played" title?"Nope! As usual I gone to see the charts first, read the text later. From the first chart I gone back to the text and started looking for the word "Nvidia" in the article.
Funny how AMD sponsored games work fine with any hardware while Nvidia sponsored games always have major stability or performance problems with AMD hardware.
JarredWalton - Wednesday, November 12, 2014 - link
I think the bigger issue is more performance and stability on AMD in general. NVIDIA has more driver engineers than AMD and they tend to work with more gaming companies; even for AMD titles (e.g. Civ:BE), there was a "Game Ready" driver available before the game shipped. AMD did better a few years back, but it seems they're falling off the pace a bit right now.siriq - Wednesday, November 12, 2014 - link
Any chance to the with different cpus? Take a look on AC Unity : http://postimg.org/image/meg9uhaxp/ AMD cpu is no go. Many games could be affacted as well. Short note: i do have AMD cpu, not one but 5.siriq - Wednesday, November 12, 2014 - link
Just hate my keyboard. it is dying. Doesn't take my press on buttons anymore.JarredWalton - Thursday, November 13, 2014 - link
Unfortunately I don't have any AMD mobo/APU to do the testing on right now, but it wouldn't surprise me to see CPU bottlenecks come into play on lower performance CPUs/APUs. The CPU bottleneck appears to be around 110-115 FPS on my 4.1GHz i7-4770K (at Low/Medium quality), so at higher quality and with dual GPUs it would probably be lower than that. I think most people would be fine with a single fast GPU and an APU, but breaking 60 FPS might require something more potent.yannigr2 - Thursday, November 13, 2014 - link
On one hand AMD is letting people go, so maybe less programmers in the drivers team. On the other hand many games are starting to be made with AMD hardware in mind thanks to the consoles. This two things go to opposite directions. If AMD in a year or two, looks better in the driver area, then it wasn't the number of engineers as big a factor as we both think, but the way games where made. Games where made on Intel+Nvidia hardware from what I remember and just tested for compatibility on AMD hardware. So Nvidia had to address less problems than AMD.Of course in the above though Ubisoft is the exception considering their tight relationship with Nvidia. They will always program games with Nvidia hardware in mind.
But what I want to say is that. In a year or two, if AMD with less stuff in the programmers team is starting to look better, then those conspiracy theories I am talking about now, could have a very large dose of reality.
FlushedBubblyJock - Thursday, November 20, 2014 - link
AMD has been letting people go for years.AMD already had the lesser of the people, because any one with a brain wanted to work for the company with BILLIONS in the bank, nVidia.
All the conspiracy theories in the world pale in comparison to the common sense dollar analysis, call it the terrible social consequences of capitalism for amd fans...
Not too mention, AMD is to be pitied, it's end user fan base is constantly demanding more for less and is the worst of the penny pinching freak jobs, the exact kind of customer and consumer every decent businessman wants to avoid 24/7/365... as they drain the viability of the company.
Scrimping little penniless tightwad complainers... always demanding a better deal - the only upside being the insane loyalty and "for the poor underdog" mindset that can be used and wielded as a powerful yet dishonest PR ax.
Well, in the end, that spells destruction, and in the mean time, it spells under performance, problems, lagging, and inferior quality in every area.
Maybe the FED can print up a multi billion dollar bailout "for free".
FlushedBubblyJock - Thursday, November 20, 2014 - link
That isn't even true, take Dragon Age and it's titles for instance.AMD is the one pulling stunts to strip the competition of performance when one of their 4 "game devs" sends off email answers to game companies.
Nvidia has literally 100 traversing the world, AMD had 4 last time I saw it publicly revealed.
So there's only so much AMD can do with 4 underpaid disgruntled and overworked stringers facing the firing ax every time the quarterly loss sheet hits the news.
Maybe the EU should make a new progressive law whereby nVidia must begin optimizing AMD GPU's as "each according to it's need" would please amd fans internationale.
Gigaplex - Wednesday, November 12, 2014 - link
"but I don't think AMD has spent as much time optimizing their drivers for the game"I'm tired of seeing a need for special application-specific "optimisations" (read: corner-cutting) in the low level drivers just to make the game work. If the games used the APIs correctly, and the drivers implemented the APIs correctly, driver optimisations should benefit all games.
FlushedBubblyJock - Thursday, November 20, 2014 - link
Oh great, then mantle should be a smashing success.TrackSmart - Wednesday, November 12, 2014 - link
Maybe it's just me, but it's hard to keep up with all of the new GPU naming schemes. For those of us who are a little behind the times, it might be helpful to reference some of the older "equivalents" in articles like this. It would let us more quickly figure out where our existing hardware is likely to fall on these charts.Doing a bit of searching, I can see that the R7 250X is essentially equivalent to a Radeon HD 7770 (and hence not far off of the older 6770 and 5770). That's useful knowledge! It says that yesterday's "mid-range" videocard is going to have a tough time handling the current generation of games, even at compromised settings.
Thanks for keeping us informed.
JarredWalton - Thursday, November 13, 2014 - link
True. The R9 280 is the same as the HD 7950, and the R9 280X is basically the same as HD 7970 (GHz edition or overclocked a bit). R9 285 is a different core, but performance should be about the same as the R9 280 as well. On the NVIDIA side, there haven't been as many straight rebadges (at least for the GPUs I have), though in terms of performance I'd estimate the 860M as being relatively close to the old GTX 580 in performance. But I could be off there, so don't quote me. ;-)FlushedBubblyJock - Thursday, November 20, 2014 - link
rebranding is no longer a topic since AMD dove into the lake of fire and lies head first in that area...I find that extremely unfortunate as all it would take is the old name of the card next to the new name of the card in the charts 7850 7870 9750 ... everyone would realize it's been like years since AMD made anything new - except for 290x hawaii which I'd assume is just a bloated out pitcairn or tahiti.
I have to admit it's amazing their cards are even on the charts - I think AMD has pulled like 30% performance out of the driver pile across the board over time - the image quality has suffered but that is never a topic anymore either because we have to save the drowning underdog, it's a worldwide group effort.
BlueScreenJunky - Thursday, November 13, 2014 - link
Nice article, waiting for AC:U as it might be my next purchase if it runs OK on PC.Could you maybe also run a few tests with either different CPUs or downclocking / disabling cores on your i7 in 1080p to see if some games may be CPU limited ?
JarredWalton - Thursday, November 13, 2014 - link
I might be willing to test a couple specific configurations once the main testing is complete, but doing a second CPU with every GPU configuration would double my testing time. Oh, and AC:U is already a pain in the butt to benchmark; there are four times of day that you might be in, and each has slightly different performance characteristics, plus the crowds are rather non-deterministic.I started testing and then after a few hours I started realizing some of my results didn't make much sense, so I had to go back to the drawing board. Now I'm having to run each test sequence a couple extra times to verify results, but I've finally got reasonably consistent numbers. There's more variance than in other games (I'd say as much as 5-10% between runs on the same hardware), but the I run each setting three times and verify I get at least two scores that are within a couple percent, then I take the highest score out of those.
More than you wanted to know, I'm sure. I'd guess AC:U is a bit less CPU limited than some games (or at least it's looking very GPU limited so far). What GPU do you have? I can at least give you a guess as to how it will run AC:U, though so far I've been testing NVIDIA GPUs.
raghu78 - Thursday, November 13, 2014 - link
JaredI would like you to test Dragon Age Inquisition when the game is out . The game looks amazing with the Frostbite 3 engine and its a well reviewed game which could find a place in your testing suite. Call of duty Advanced Warfare is worth testing as the game is being well reviewed and is a popular game. Surprisingly COD AW is being praised for its improved graphics. Finally Farcry 4 is the last game from this holiday season which is worth testing as it will surely be a great looking game and a worthy GPU benchmark title.
raghu78 - Thursday, November 13, 2014 - link
btw how is the R9 290X doing so poorly in your review when other reviews show it performing on par with GTX 970 and GTX 780 OC and GTX Titan. Have you picked a corner case in this game where AMD cards do poorly ?http://gamegpu.ru/action-/-fps-/-tps/lords-of-the-...
http://www.pcgameshardware.de/Lords-of-the-Fallen-...
http://www.benchmark.pl/testy_i_recenzje/lords-of-...
JarredWalton - Thursday, November 13, 2014 - link
Obviously I don't know what areas other places are using for testing, but I did choose a place where I noticed more of a slowdown -- basically, I look for "worst case" areas for benchmarking, as people usually want to know how things will fare in the most demanding areas. Running round indoors, LotF has much better performance, while outdoors there are some vistas that you encounter where frame rates can take a dive.If they're getting minimum FPS relatively close to the average, obviously they're not hitting some of the more demanding areas. I didn't bother to try any translation services, so I'm not sure what exactly they're saying or how they're testing, so that's about all I can say.
One thing somewhat interesting to note is that the patch that improved performance on GPUs with 2GB and less RAM actually tended to drop performance a bit on the R9 290X. At 2560x1440 I measured about a 5FPS drop on 290X, while only a ~2FPS drop on 980/970. 1920x1080 Ultra showed a similar drop on 290X. I have no idea why that happened, but it's very much reproducible.
FlushedBubblyJock - Thursday, November 20, 2014 - link
I went to your gamegpu link and saw 290X doing just as poorly.raghu78 - Thursday, November 13, 2014 - link
JaredSorry missed out Middle Earth Shadow of Mordor which is one of the best games of 2014 and is graphically quite demanding. It would be a game worth benchmarking.
JarredWalton - Thursday, November 13, 2014 - link
I do have some benchmark numbers from the game, but there's a lot of weird stuff with the Shadow of Mordor benchmark. For one, it includes time from before the benchmark starts, so that throws things off. The minimum FPS is also pretty much useless, so you have to use FRAPS to get a better idea of a "true" minimum. At this point it's older so I have other games to test, but maybe a short article would be worth posting if I get a chance.FlushedBubblyJock - Thursday, November 20, 2014 - link
Isn't Dragon Age Inquisition an AMD biased Gaming Evolved game ?All their dragon age gaming evolved games violated the spirit of the amd gamers manifesto and crippled nVidia cards.
jann5s - Thursday, November 13, 2014 - link
The comparison of the various quality settings including the visual impact of the graphics and the performance impact was very nice. I would love to see more along these lines.eanazag - Thursday, November 13, 2014 - link
I have been reading these and appreciate them. I have two R9 290's and I can tell where they'd likely fall if I were to pick up this game. My CPU will cause me to lose some FPS and I'm not crossfiring the cards as my current setup causes them to run really hot in the case (non-blower model).I have found that I'd like to see double the spacing of PCIe x16 slots for better thermals in a 2 way SLI/Crossfire setup from mobo makers. One card plays fine and the other goes to $#!+ on temps to me - ~10+ Celsius higher when pegged out.
FlushedBubblyJock - Thursday, November 20, 2014 - link
Ok. 1080P ultra 280x 48 290x 63, soooooo 290 is like 55FPS.Yeah, you should be able to tell, but it's getting worse for accuracy with the cpu offloading, and you didn't say what your cpu is... but just see any 290 bench game w similar cpu reliance , and you have - reduce percentage accordingly.
ezschemi - Thursday, November 13, 2014 - link
I really really enjoy these short articles (short compared to the reviews).What I would like to see
* a GTX 970 SLI configuration as these are pretty reasonably priced
Games to Test
* Dragon Age Inquisition with and without Mantle
* Assassin's Creed Unity
theharleyquin - Wednesday, November 19, 2014 - link
No GTX 760 *tear*. Must be time to upgrade.