Home / Component / Graphics / Alleged benchmark results of AMD Radeon R9 300-series ‘Fiji XT’ published

Alleged benchmark results of AMD Radeon R9 300-series ‘Fiji XT’ published

A Chinese web-site that leaks unofficial information about future central processing units and graphics processing units has published what it claims to be the first benchmark results of AMD’s next-generation Radeon R9 300-series graphics card code-named “Fiji XT”. If the benchmark results are correct, then the graphics adapters will offer rather high performance in all existing games.

Chiphell, which has a reputation of leaking both accurate and inaccurate information, has published two charts representing average performance and power consumption of AMD Radeon R9 300-series graphics card – which was called “Captain Jack” graphics adapter to emphasize that it belongs to the Pirate Islands family of products – across a number of benchmarks versus current-generation graphics boards.

amd_radeon_artwork_angle_new

The list of benchmarks that were used to get “average” FPS [frames per second] number features a long list of games, including Alien Isolation, Assassins Creed: Unity, Battlefield 4. Crysis 3, Dragon Age: Inquisition, Far Cry 4, Metro Last Light Redux and many others. All titles were benchmarked in 2560*1440 resolution with all eye candy (antialiasing, level of details, etc.) set at maximum. The testbed used was powered by Intel Core i7-4790 (3.60GHz) microprocessor. The testers claim that they utilised AMD Catalyst 14.11.2 beta and Nvidia GeForce drivers 344.75.

While the results clearly show that the Radeon R9 300-series “Fiji XT” graphics card is a clear winner that provides around 10fps performance advantage – or a solid 16 per cent – on average versus a factory-overclocked GeForce GTX 980 graphics board, it should be noted that the test conditions are not clear and thus the result cannot by verified anyhow. Moreover, even if the result is correct, then the Radeon R9 300-series “Fiji XT” is only 30.9 per cent faster than the Radeon R9 290X “Tahiti XT”, a not very high generation-to-generation improvement.

amd_radeon_r9_390x_alleged_benchmark_average_fps

At present the Radeon R9 300-series “Fiji XT” (or Captain Jack) graphics cards are probably already available in test labs of AMD’s partners in various southeast Asian countries, therefore, it is possible that an enthusiast employed by a partner could benchmark one of them. What should be noted is that at this time nobody has final specifications of the product that will be sold as the Radeon R9 390X. The GPU configuration is unclear: it may feature 4096 stream processors and 256 texture units, but AMD can well disable two compute units inside its chip for redundancy reasons and the final chip will sport 3968 SPs and 248 TUs. Even if the configuration of the chip itself is locked (i.e., will not be changed), both AMD and Nvidia finalize clock-rates weeks, not months, before commercial launch. As a result, whatever it was used to get benchmark results, it was not a product that will be known as the Radeon R9 390X. Finally, while the Radeon R9 300-series “Fiji XT” is believed to be based on the GCN 1.2 architecture and thus should have decent software support now, the drivers used by the benchmarkers were re not final drivers for a shipping product.

amd_radeon_r9_390x_alleged_benchmark_power

It is interesting to note that if the performance and power consumption measurements are correct, then the Radeon R9 300-series “Fiji” graphics cards should be rather power efficient. Keeping in mind that AMD’s Radeon R9 285 “Tonga Pro” consumes lower amount of power than AMD’s Radeon R9 280 “Tahiti Pro” (both chips are made using 28nm process technology and feature similar amount of transistors), increased power efficiency of GCN 1.2-based “Fiji” GPU may be true.

AMD did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: We have no idea whether the benchmark results are correct or not, but even if they are, “average” fps across multiple games is a metric that is not good. Moreover, assuming that the results are correct, if the Radeon R9 300-series “Fiji XT” graphics card (we assume that the anonymous tester used a higher-end model) is only 15.9 per cent faster than a factory-overclocked GeForce GTX 980, then it will hardly be a performance breakthrough, despite of massive amount of stream processors and HBM DRAM memory with incredible bandwidth.

Become a Patron!

Check Also

Intel’s Xe3 GPU architecture is already complete, successor already in the works

In a recent podcast interview, Arc Graphics leader and Intel Fellow Tom Petersen revealed some …

66 comments

  1. In the chinese site it was said that this GPU was 380X and not 390X.
    Do you Guys really thinks 16% improvement over a factory overclocked 980 using a unfinished product a bad result ? TITAN II is espected to be how much faster ?

  2. It’s better be 380x, 390x should be way higher than this *fingercrossed*

  3. Laughed quite hard at this

    “Moreover, even if the result is correct, then the Radeon R9 300-series “Fiji XT” is only 30.9 per cent faster than the Radeon R9 290X “Tahiti XT”, a not very high generation-to-generation improvement.”

    GTX 980 V 780Ti 7% improvement??? Captain Jack 30% faster from an Engineering Sample, whilst at it taking the Perf/Watt crown.

  4. While not very high, people gotta remember, we are reaching silicons limits. 😛

  5. This is just an ES sample still have 20-40% performance under hud so no need to cry on other side like in fp64 compute GCN 1 was way way better than kepler surly this will happen again as maxwell is based on performance per watt and GCN is on sheer performance.
    Simple advice to amd – u have only 1 year to get limelight coz if u dont use HBM than nvidia will use that in late 2015 or in early 2016. So just put all crap aside use HBM and enjoy limelight for whole year

  6. Wow that’s 10% better power efficiency then the GTX980 very surprising result. I was expecting a 290W sucking monster not a 197W card. It dose use a little more then the GTX980 but it’s got a good deal more FPS for a tiny power bump.

    With that level of power draw I’d be expecting another card a XTX or something similar which gives a bigger jump down the road…

  7. Since AMD mentioned about stock liquid cooling on there top end GPU, this is a GPU of under 200W TDP and there is no point in using a liquid cooler on this. The GPU here is probably a 380x which easily competes with GTX980. R9 390x is something really really big with a die size of 500-600 mm2 that really needed a liquid cooler. It seems that AMD is finally starting to retaliate.

  8. Most probably this is an upper mid range 380x GPU

  9. 30% Not a big performance gain……If it was a pay rise I think the author may change his mind

  10. Charles Charalambous

    It was a lot more than a 7% improvement, if this benchmark comes true though it is finally a step in the right direction for AMD but seriously get your head out of your AMD fanboy ass. It was a lot more than 7%. More the 15-20% Mark and given how powerful the 780ti was that was impressive, and the energy consumption and temps and noise levels went down insanely. What the OP probably meant is 30% given how much AMD needed to come back, especially when nvidia is releasing at least 1 new GPU next year. It is not really enough. Anyone who is not a fanboy of either side will not jump to AMD for this If they have a 670/760 SLI setup, 780TI or 900 series card.

    Also a light OC on my 970 gives me 980 level Performance while not going over 70c. Over people have got a lot more. Add in a 980 OC overclock and it basically performs the same as this new AMD card if rumors are true. Nothing to write home about. It will just be better price to performance ratio but nothing to make people jump ship.

  11. Fanboy ass? No, just the ability to read reviews.

    “In terms of pure performance, we find the GeForce GTX 980 exceeding the GeForce GTX 780 Ti by 7%, which doesn’t look huge, but consider that the GM204 has nearly 2 billion fewer transistors and about a 50% lower TDP, and something tells us that the GM204 won’t be the biggest chip NVIDIA will design on this architecture. Compared to AMD’s Radeon R9 290X, we see a large performance gap of up to 20% at 1920×1080, although it shrinks to 10% at higher resolutions. This makes the GTX 980 an excellent choice for beyond-HD gaming, but you should still have two cards running in SLI for 4K.”

    http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_980/31.html

  12. Charles Charalambous

    It was more than 7% a friend has got 20% over his 780ti. DEAL WITH IT

  13. Does your friend publish benchmarks? If not then why would we care about that as opposed to the documented evidence? You can claim that the power draw along with performance makes it more impressive than just the performance increase, but you can’t simply say: I SAW MORE THAN 7% SO THERE.

  14. Technically 8% but it’s pretty much the same. Also, should keep in mind that this card is on the 20nm node and is using HBM, which uses 32% less power than GDDR5. If it was 28nm using GDDR5, it would be in the 220-240W range. Of course, that’s still a massive improvement over their previous cards. There’s also the fact that this card is the 380X and not the 390X, which makes me wonder what the performance on the 390X is going to be, and how that’s going to stack up against GM200.

  15. Silicon limits are going to be hit around 8nm…

  16. This card is the 380X, not the 390X. It wasn’t specified in the benchmarks, and I’m not sure why the author assumed it.

  17. let’s hope it, I wanted to buy a 970 but now I’m better off waiting for AMDs turn

  18. haha now who’s the fanboy. I show you facts and you pull some fictional friend story out your ass to backup your null argument.

    Getting back to my original comment, Nvidia made a 7% jump in performance whilst saving around 50% on power, that was deemed as impressive, which it is. My point is, if these leaks are true, then AMD has went from a card using 295W whilst gaming to 197W as stated above in the graphs, which is around 50% also, but AMD started with a higher TDP to start off with, so that 50% is around 100W not 80W of the 980 over 780Ti. Yet whilst saving more power, they have managed a 30% increase over last gen. I’m sorry but, this is not being a fanboy, just looking at the information in front of me and realizing that AMD had a bigger task to achieve these numbers than Nvidia.

  19. Yo genius at least watch out some review!!!
    http://www.techpowerup.com/reviews/ASUS/GTX_980_STRIX_OC/25.html

    Custom 980 vs reference 780ti—-> 10% difference, plz if you don’t know something check it out or you gonna make a big fail man. What a nice fanboy :):)

    Editi:

    Oh yes they can say that the 980 consume less… Spending 3, or less, bucks per months will make me billionaire xD

  20. Charles Charalambous

    KitGuru stated more of a gain but you keep ignoring their benchmarks because it prooves you otherwise. When you start paying for your own electricity it is a big deal. Gains are 10-15% with a 980. After the very easy OC with hardly any voltage adjustment that goes up to 20-25%. Making it on par if not better than this new AMD card which is a 30% gain compared to the previous model. Despite AMD having all this new tech in there. Blidn fanboys will be blind. I scored myself bettter than 780TI performance for £280. I am happy. My point is this card is not goign to make anyone swap for an upgrade. Techpowerup really underquoted the perfomance of the 980 which has only got better with drivers.

    We also don’t know if this benchmark shows the stock clock of the new card or an OC to make it look better than what it is. Either away it’s not a game changed with AMD’s new tech. Which says everything about AMD at the moment really.

    I want them to make a game changer as competition is needed in any industry, this is underwhelming to say the least

  21. I saw on some rumors sites that 390X could still use 28nm and not 20nm. Do you really think it is going to be in 20nm already?

  22. in the chinese forum they said “midrange” and 380X, assuming the bench is true ( if not there is not a point debating around it) we have to accept it is the 380X till new rumors say the opposite.

    ( sorry my rusty english kkk)

  23. Well no, of course it’s not going to make people run to buy one. It’s the 380X, not the 390X.

  24. It should be, especially since their CEO said their 2015 products would be on 20nm, but we don’t really know for sure. If they made this improvement while staying on 28nm then it would be insane.

  25. Even the graph above has the 980 7% above the 780Ti….

    Okay lets pull some benchmarks from Kitguru

    http://www.kitguru.net/wp-content/uploads/2014/09/metro-1600p.png

    Even heavily overclocked its still only 13% faster and 3% at stock. Some other benches will vary, and this is probably one of the worst showings for the 980, but other games still don’t show that much difference. A point to make i suppose is they are using a 780Ti Direct CU II, but that is only mildly OC, and your still only looking at around 4% on this game Stock V Stock.

    This is Kitgurus words, read the brackets.

    “At reference clock speeds the GTX980 is very closely matched against the overclocked Asus GTX780Ti Direct CU II OC which we included in our review today. The GTX980 in our testing has a (slight performance edge) most of the time – while demanding 30% less power to do so. Those who have complained about the GTX780Ti having only 3GB of memory will be pleased that the GTX980 now has 4GB – useful when dealing with heavy texture based titles across multiple screens or at 4k resolution.”

    “KitGuru stated more of a gain but you keep ignoring their benchmarks because it prooves you otherwise. When you start paying for your own electricity it is a big deal. Gains are 10-15% with a 980. After the very easy OC with hardly any voltage adjustment that goes up to 20-25%.”

    So please show us where this claim Kitguru made about it being more, also be free to show the benchmarks where this 10-15% gain is at stock and 20-25% gain after OC.

  26. But still it a big improvement over my Pentium G 2020 lol…

  27. Charles Charalambous

    Sure no problem wait until I get home and get you the figures butthurt amd brainwashed sheep

  28. How very mature, you know people who resort to insults usually mean they have no argument.

  29. The first sign of someone being a fanboy is when they call someone else a fanboy for apparently no other reason besides disagreeing with them. You don’t seem to realize that the benchmarks done by tech sites all indicate the GTX 980 gives about a 15% performance increase at best over a heavily OCd 780 Ti when it’s heavily OCd. The average advantage is closer to 10%. This is your word against the word of every tech site that has ran benchmarks on the GTX 980. What happened was your “friend” got a card from the tip of the silicon pyramid. That in no way represents the results most people are most likely to have.

  30. I read an article somewhere that a OC 780Ti vs OC 980 is very close and trade blows, but are within 5% of each other after both been overclocked to the max.

    EDIT: Here you go found it

    http://alienbabeltech.com/main/overclocking-max-gtx-980-vs-r9-290x-vs-gtx-780-ti/3/

    http://i1.wp.com/alienbabeltech.com/main/wp-content/uploads/Untitled-18.jpg

    Bottom 10 games (Newest DX11) @ 1600p

    780Ti OC AVG 57.88FPS

    980 OC AVG 59.44FPS

    R9 290X OC AVG 50.3FPS

  31. GrimmReaper WithaSpoon

    oh the irony.
    get yo shit straight,then come back.

  32. GrimmReaper WithaSpoon

    finally, i can’t wait.
    and if that is true, I really want to see the exact specs of the GPU->memory,bus width,bandwidth etc.

  33. Charles Charalambous

    http://www.techpowerup.com/reviews/MSI/GTX_980_Gaming/16.html

    MSi Model, which yes comes with a little oc but nothing like what people are doign when they get these cards.

    At least another 5% gained on a reference 980. so that makes a total of 10%+ (dependign on resolution etc) faster than the 780TI. Add in an OC that can go up to 25-30 if lucky but being conservative that is 20%.

    Amd have just put in some impressive tech including memory bandwidth and with a 30% performance gain over it’s previous generation it is barely ahead of a 980. With more cards to follow. So ANY 980 You buy that is MSI, ASUS etc will actually be at least 5% faster than a reference model (Which no one buys face the facts). Making it underwhelming no matter how you cut it, apart from they have finally become energy efficient. In all honesty I am fed up with there being 2 brands. I want someone else to enter the GPU arena and give both AMD and Nvidia who both have their head too far up their on arse imo.

  34. No son you are WRONG. you should have 2 290x crossfire for 4k, or 8k. I completely disagree with the gtx 980 and it’s 256bit memory bus being to handle 4k better than my crossfire setup.. I’m sure it’s close for a next gen card. The gtx 980 wasn’t enough for ME to upgrade my gtx 780ti sli setup either.

  35. So now we’re comparing OC cards, we have been talking about Stock V Stock. Look at the chart below, when both 780Ti and 980 have been OC the 980 is 2.7% ahead over 10 of the latest games.

    http://i1.wp.com/alienbabeltech.com/main/wp-content/uploads/Untitled-18.jpg

    Bottom 10 games (Newest DX11) @ 1600p

    780Ti OC AVG 57.88FPS

    980 OC AVG 59.44FPS

    R9 290X OC AVG 50.3FPS

  36. Agreed!

    A card which consumes under 200w does not warrant liquid cooling(which is supposed to be used on r9 390x), therefore this is most probably r9 380x, or an underclocked engineering sample of r9 390/390x.

    However, that doesn’t really matter. It’s obvious AMD reduced power consumption by a large margin(by implementing the GCN 1.2 in conjunction with HBM), and since this card consumes under 200w, they have sufficient room to introduce a ~270-290w card, AMD will be able to counter any threat from Nvidia.

    That said, If I had to guess which card this is, I’d say it’s r9 380x.

  37. What exactly am I wrong about? I haven’t mentioned 290X and nor have i mentioned 980 for 4k gaming. I’m talking about the 980 vs 780Ti being very close in performance…

  38. I agree with you that AMD has an advantage over Nvidia, in a sense that they’ll get access to HBM much sooner than them. That said, if you really expect AMD to sit idly during that time, and do nothing, you are mistaken. They will be busy working on a true successor to GCN 1(1.1 & 1.2 were incremental updates), which will most probably debut in 2016.

    Let’s be realistic here – Neither Nvidia, nor AMD will sit idly and do nothing. They’ll do everything they can to surpass each other. Basically, they’ll be trading blows, as they always have. If Nvidia takes the crown, AMD will shortly retake it after that, and vice-versa. If only they gave Intel this sort of competition…Hopefully Zen will rectify that.

  39. Charles Charalambous

    Different results to techpower up and older drivers. 980 has been proven to be better than a 780 and once you OC it GG WP. AMD got a 30% Boost and are only a tiny bit ahead. DISSAPOINTING END OF

  40. http://www.guru3d.com/articles_pages/dragon_age_inquisition_vga_graphics_performance_benchmark_review,7.html
    here’s another example.

  41. sorry man.. wrong reply

  42. An example of what? I’m sorry but I have no idea what you’re talking about. We’re talking about the above benchmarks, and if true how well AMD done to achieve them. I’m sorry but I’m bewildered to what your actual point is.

  43. Okay that makes sense, I’ve done that myself, clicked reply to the wrong person.

  44. This is my last reply on this subject. As you can see, once overclocked the 980 has a very little lead. This pretty much puts this to bed, both tested in the same system, only difference is the 980 has newer drivers.

    http://www.techpowerup.com/reviews/MSI/GTX_980_Gaming/28.html

    http://tpucdn.com/reviews/MSI/GTX_980_Gaming/images/perf_oc.gif

    http://www.techpowerup.com/reviews/EVGA/GTX_780_Ti_Classified/27.html

    http://tpucdn.com/reviews/EVGA/GTX_780_Ti_Classified/images/perf_oc.gif

  45. You should be comparing the 980 to the 780, not the 780Ti, otherwise you’re comparing cards from different performance brackets (it would be like comparing the 7850 against the 6950).

  46. This is just a naming scheme, until they bring out a Ti or Titan 2, this is their flagship card from this Gen compared to last gen.

  47. Charles Charalambous

    I’ve seen bigger gaps but whatever. Ultimately you are saying that Nvidia 900 series did not have hardly any performance increase yet is still only 10% away without being OCed to AMDs new offering. GG WP

  48. It should have 4096 SPs (256 TMUs, 64 CUs) and a 4096-bit memory bus running at least 1.25GHz giving 640GB/s of bandwidth (1GHz = 512GB/s). That’s what we know about the 390X, the 380X we still know absolutely nothing.

  49. GrimmReaper WithaSpoon

    whoever wrote that,missed one thing.
    it’s not 4096 Bit memory bus, it’s the amount of VRAM.

  50. It’s also the bus width because each 1GB stack has a 1024-bit bus, and there should be four stacks.

  51. not a fan boy of amd because of their ccc, its terrible so it makes me favour nvidia. even though I am using amd right now and im intruiged by the new hmb hardware. Im reading fanboy comments from NVidia acting all stupid about the performance not that great improvement….

    who even said that the sample was 390X. it could actually be there lowest r9 370. I doubt they would go all out testing their best. and if im right….oh man! NVidia are in trouble. at least for a year.

  52. It’s most likely the 380X, just going by the performance and power consumption.

  53. The big thing I’m looking at is gk104 to gm204. 980s are for 680 upgrades, whatever gm110 card launches is what the 780 series owners are waiting on.

  54. so you think 🙂

  55. Kristijan Vragović

    @Steve Smith… Can you tell me based on these leaks what you think about core count in this gpu?? I believe that is lower than predicted, around same number as 290x or lower. Also i think IF ANY OF THIS IS TRUE that they succeded to improve efficiency of GCN cores for gaming because that was not the case before… Just wanna see what your thoughts are

  56. Kristijan Vragović

    You just don’t get it yourself what you typed… When the R9 380x, because that this model is if the charts are accurate, goes out will also be overclocked and people will overclock them further. Also, if this is accurate, this is early version and will be even faster closing the revealing date… So i just can not understand what your problem is… And R9 290X is one year old product which holds up pretty good to a 970. AMD and Nvidia trading blows every year, and every time someone comes at top. R9 were great card when presented and was great in games, but Nvidia answered like it was before and will be in the future.

  57. Kristijan Vragović

    I saw different answer when they were asked about manufacturing process… Many sites confirmed that 20nm for gpu will not happen

  58. It’s hard to tell really until we know if this is on a new node, efficiency gains etc. But if I was to take a guess, then I would agree with you. I do think this is the 380X and I would say it probably has around the same core count as 290X, but with improved efficiency on the GCN cores and HBM. All helping with power and performance.

  59. Kristijan Vragović

    Well, i’ve read a few articles saying that a 20nm nodes for a GPUs are no go… So i have based my prediction on that.
    http://fudzilla.com/home/item/36433-globalfoundries-got-ex-amd-svp-of-operations

    This article is one thing that makes me more assure of that. We all know that all of this are only predictions and rumors until confirmed or dissmised but, AMD is cooking Zen, and K12, and they need a opponent to a Nvidias Pascal in the end of 2015. Since Carrizo is made by making the cores more dense i think that they have made the same with GPUs… That’s what made consumption lower allong with HBM, and like you typed GCN improved efficiency whille gaming…All in all 2015 will be a interesting year. Hope that GloFo will finally deliver AMD node they need, and since, atleast thats what rumors suggest, they are top customer AMD could finally deliver… Now we have to wait to see what is true and what is fake.
    If AMD finally awakes then all of manufacturers will be at last forced to be honest about performance and price, and that makes me happy,

  60. i don’t think the 980 was supposed to replace the 780 Ti. nvidia is holding back the 980 ti, 990, whatever they will call it, so the die hard fanatics would buy the 980 then buy whatever comes after it again a few months later. they are holding back the top cards to milk their user base.

  61. I was reading your comments and I have to say….you sound like little kid who lost argument.

  62. Titan 2 should be 25-30% faster than the 980, if the difference between the 680 and 780 (8SMX vs. 12SMX, 256-bit vs. 384-bit) was any indication, as it’s supposed to have 50% more CUDA cores and a 50% wider bus. If this benchmark is for the 380X then the Titan 2 doesn’t stand a chance against the 390X. I’m fairly certain that this is an under clocked 390X, probably running at 800MHz on the GPU and 400MHz (204.8GB/s) on the memory, if the benchmarks in Sisoft’s database are an indication of these results.

  63. Well, not to be an AMD fanboy, assuming these data are correct, AMD is gonna kick some serious ass with their new generation of cards.
    If this is the 390x, Nvidia would be slightly screwed, the 390x should have similar price as a 980, but with faster performance and better efficiency, then Nvidia will have to have some sort of price cut, or else any sane gamers would choose amd over their offers. However, this possibility is unlikely, since putting their flagship at a mere 197 Watts doesn’t seem like an AMD thing to do. Besides, for a card with a 197 Watt power consumption, with very good efficiency, liquid cooling is very unnecessary, so the liquid cooled card is probably an upper 200 watt one, likely to be the 390x.
    If this is the 380x, then Nvidia would be seriously screwed, a card that costs around $150-200 less will have better performance and better efficiency, and the 390x, with about the same price as a gtx980, is gonna have like 60% higher performance, and better efficiency. And with this much performance, you will be only looking at gaming at high res, which AMD has an even bigger advantage. If that is the case, even Nvidia softcore fanboys would switch side to AMD, leave alone sane gamers who choose sides based on their actual performance and efficiency. Nvidia would need a huge price cut, talking like $150 at least, to keep themselves in the game, assuming they don’t want to just get money from fanboys.

  64. Same here.

  65. IF this is true, and that is debatable, it is not the 390 but actually the 380 GPU that this bench is about. If the rumors are true and this series uses 20NM (which is almost sure) and HBM the argument about OC ability will be void. 2015 is shaping up to be great for team RED, gamers rejoice!!!

  66. They are also not considering the prices. If this really is the 280x replacement, it will probably cost about $300 give or take – the same or less money than a $350 970 with better performance than a $600 980. That is going to to force Nvidia to drop their prices the next day after it is released.