Just be sure you case has room for 2 radiators or if your are using water cooling on the CPU, 3 radiators. Also be sure the game you want to play is supported by crossfire. For now I'd say get the 980 Ti the fury x may scale a little better but I doubt you'd notice it in actual game play and you know the SLI drivers will work. There is no way to know when or if the Crossfire drivers will ever be fixed. I remember one year AMD released it's top of the line card in December and the Crossfire drivers didn't come out until August and they still needed work. This driver issue goes back to the earlier ATI days why don't they ever address it? I still have nightmares about the ATI x200 drivers with my phone ringing non stop with customers complaining..
- Status
Just be sure you case has room for 2 radiators or if your are using water cooling on the CPU, 3 radiators. Also be sure the game you want to play is supported by crossfire. For now I'd say get the 980 Ti the fury x may scale a little better but I doubt you'd notice it in actual game play and you know the SLI drivers will work. There is no way to know when or if the Crossfire drivers will ever be fixed. I remember one year AMD released it's top of the line card in December and the Crossfire drivers didn't come out until August and they still needed work. This driver issue goes back to the earlier ATI days why don't they ever address it? I still have nightmares about the ATI x200 drivers with my phone ringing non stop with customers complaining..
esrever
Splendid
- Dec 27, 2011
- 5,391
- 0
- 26,160
I can't really say I would recommend an nvidia card below the 980ti at this point unless you really cared about efficiency. AMD cards are just faster for the same price for the price points below that and seem to have a slightly better future in dx12.
MotherBrain
Reputable
- May 7, 2015
- 10
- 0
- 4,510
Because, you know, we might want the option to buy an el cheepo case to offset our $1200 budget for graphics cards....
[@@]
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
"Now, the Nano is a special case in that it’s slower, but specifically intended for small form factor applications. If you need it, you need it. If not, there’s no reason to pay a premium for a card that underperforms the GeForce GTX 980 (vanilla, not Ti)."
It depends what you mean by underperforming. The MSI GTX 980 Gaming 4G is a mildly overclocked model (about 3% better performance than stock) which was the basis for the Nano review comparison. Whilst it's true that the 980 wins at 1080p, the spoils are shared at 1440p; a stock 980 would be a little behind at 1440p, and I doubt you'd want to run a $650 card at 1080p anyway. Energy usage between the MSI GTX 980 Gaming 4G and the Nano would also be very close. I would definitely agree that a 980 makes more sense in a system where space isn't at a premium. These cards aren't meant to compete with each other directly.
It depends what you mean by underperforming. The MSI GTX 980 Gaming 4G is a mildly overclocked model (about 3% better performance than stock) which was the basis for the Nano review comparison. Whilst it's true that the 980 wins at 1080p, the spoils are shared at 1440p; a stock 980 would be a little behind at 1440p, and I doubt you'd want to run a $650 card at 1080p anyway. Energy usage between the MSI GTX 980 Gaming 4G and the Nano would also be very close. I would definitely agree that a 980 makes more sense in a system where space isn't at a premium. These cards aren't meant to compete with each other directly.
- Jul 15, 2014
- 22,630
- 8
- 60,960
Mike Stewart
Honorable
- Jun 12, 2013
- 55
- 0
- 10,640
Truly disappointed with hierarchy chart Tomshardware !!!!
Not only they forgot to include 950gtx but they also made HUGE mistakes...
390X is WAY fatser than 980 ! in most of the games, on par with only a few games and slower only to those gimped gameworks games made for nvidia only !
and look at 390 and 290X they sit same place with 970 and 780ti.. this is a joke ! for real !
290X and especially 390 will destroy any 970 in all of First person shooters and not only FPS.. but i get it now ! tomshardware must be paid by nvidia,,, there's no other logical explanation.
Not only they forgot to include 950gtx but they also made HUGE mistakes...
390X is WAY fatser than 980 ! in most of the games, on par with only a few games and slower only to those gimped gameworks games made for nvidia only !
and look at 390 and 290X they sit same place with 970 and 780ti.. this is a joke ! for real !
290X and especially 390 will destroy any 970 in all of First person shooters and not only FPS.. but i get it now ! tomshardware must be paid by nvidia,,, there's no other logical explanation.
CptBarbossa
Honorable
- Jan 10, 2014
- 401
- 0
- 10,860
In that price range I would go with the r9 290. It is still a powerful card and can be had at the $240 price range.
johnlocke32
Reputable
- Oct 17, 2015
- 1
- 0
- 4,510
http://www.videocardbenchmark.net/high_end_gpus.html
Take this research with a grain of salt because its benchmarking after all. For more information on this research click on the name of the GPU and it explains in more detail what the testers analyzed. Many people say that this or that GPU performs better, but numbers typically don't lie. It appears this website updates all of its GPU tests daily and prices per the 1st of the month.
Take this research with a grain of salt because its benchmarking after all. For more information on this research click on the name of the GPU and it explains in more detail what the testers analyzed. Many people say that this or that GPU performs better, but numbers typically don't lie. It appears this website updates all of its GPU tests daily and prices per the 1st of the month.
CptBarbossa
Honorable
- Jan 10, 2014
- 401
- 0
- 10,860
eh.... try actually findin a gtx 780ti for $385. Heck try finding one at all. Those numbers aren't accurate.
- Oct 30, 2004
- 32,255
- 14
- 99,960
Another ranking which totally ignores the relative overclock ability of cards. When choosing a high end GFX card what % of the THG audience doesn't fire up Afterburner or it's equivalent immediately after installing ?
When trying to decide what premium tier product is "for you", doesn't OC headroom come into play ? I would think that the fact that one top tier product OCs 31% and the other 5% would have an impact on most people's buying decision. Without this information on how most will actually use the card, I really don't have a reason to read it. If CPUs were evaluated similarly, the 4690 and 4690k would also be ranked equally.
When trying to decide what premium tier product is "for you", doesn't OC headroom come into play ? I would think that the fact that one top tier product OCs 31% and the other 5% would have an impact on most people's buying decision. Without this information on how most will actually use the card, I really don't have a reason to read it. If CPUs were evaluated similarly, the 4690 and 4690k would also be ranked equally.
CptBarbossa
Honorable
- Jan 10, 2014
- 401
- 0
- 10,860
The reason you cannot use overclocking as a measurement because overclocking is a crapshoot. Some cards TEND to overclock better than others, but there are no guarantees. You cannot base a purchase on an overclock that you may not achieve.
- Oct 30, 2004
- 32,255
- 14
- 99,960
Red Herring (Def) = "something, especially a clue, that is or is intended to be misleading or distracting"
There are threads on overclocking a 4670k and if you are looking for 4.8 Ghz, your definition might be applicable. But look at OC threads on OC oriented sites like overclock.net and you see spreadsheet compilations of member overclocks numbering in the 100s... I don't see anyone with adequate cooling not getting to 4.4 or 4.5 Ghz. So if we are seeing OC's on 4670k's of 13% to 26%, that doesn't mean another CPU doing 4 - 8% would be anywhere in the same ballpark.
After you read 18 reviews on a particular card and every one of them gets between a 26% and 31% over "stock", while another never does better than 4 - 8%, I'd say that's far from a crapshoot. The overall market might be seeing 24 - 33% versus 2 - 10% but that doesn't mean you'll get even odds that that any particular top tier 9xx series wouldn't overclock better than any top tier 3xx series.
By my definition, those are odds I'd take at any casino. That's why they have spreads in football betting. Because it is certainly not a "crapshoot" when a team like Ohio State plays a team like Rutgers... we see that, when tested, OSU is 6-0 and Rutgers is 2-3 and tho any team can lose on any given day, that doesn't make it a "crapshoot". The betting line is not about who will win the game but just by how much.... the two teams are just not in the same 'class of performance". So the "crapshoot" when comparing the two top GFX cards, is not about who wins, just about by how much.
When doing our own or user builds, we base the purchasing decisions on how each custom PCB is built and whether it has proven via published testing that the performance reflects what is expected from the componentry used. The "crapshoot" as you call it, or what mathematicians call the "standard deviation", means that if you have 15 samples test between 25% and 31%, then the variation in the general piopulation might range from 22% to 33%. In no way will the card that OC's only 4-8%, ever ever get up to thew 20% range. The fact is, the R9 series is very aggressively overclocked 'in the box", leaving very little headroom. After all, the 7xxx series were tweaked a bit to make the R9 series and the R9 200 series was tweaked some more to make the R9 300 series. There's simply just no more room to be had. And the new Fury series has less OC headroom than the R9s.
When you look even at the performance of the same card from various vendors, you see trends. For example on the 970s, the Gigabyte G1 wins almost every head to head comparison. Both MSI and Gigabyte routinely break 1500 Mhz boost clock, whereas the competition from other Big 4 vendors fails to break that barrier in just about every review. With our 7xx based builds, the lowest OC experienced was 22%, highest was 28% ... with R9 200 series, the highest was 16 % and that single build stood well above all the others
So rather than a "crapshoot", the data shows an "overwhelming probability of result". Techpowerup recorded a 26% OC (1090 / 863) over reference speed in their OC review of the Asus 780. On the box I am typing from, (twin water cooled 780's in SLI), I managed 1078 (25%). On a subsequent MSI 780 build, we got 1104 MHz (28%).
Every piece of silicon is different, but there is a determinable statistical range which can reasonably be expected. The reference 980 Ti has scored 102.6 fps in BF3 as tested by techpowerup showed. Aftermarket cards hit:
Palit GTX 980 Ti JetStream = 133.1 fps (30% OC)
ASUS GTX 980 Ti STRIX = 131.7 (28% OC)
ZOTAC GTX 980 Ti AMP! = 130.4 (27% OC)
MSI GTX 980 Ti Gaming = 130.5 (27% OC)
Gigabyte GTX 980 Ti G1 Gaming = 134.8 (31% OC)
EVGA GTX 980 Ti SC+ = 126.6 (23% OC)
That is an average of 1.28 .... 1.29 if you eliminate the EVGA card which is logical given that it's not really a non-reference card ... it's a non-reference cooler on a reference PCB.
The standard deviation for the true non-reference cards is just 1.8% .... not exactly what you could call a "crapshoot".
OTOH, the 390x results are quite different. The reference 390x scored 84.3 fps in BF3 as tested by techpowerup. The MSI 390x managed just 7% more. It think that we can expect MSI to put the same effort into making the 390x non-reference card as they did their 980 Ti and yet one managed a 31% OC and one managed just a 7% OC. Given the consistency of test results and standard deviation shown above, the only conclusion that can be drawn from the above that the difference is related to the GPU.
Again like the previous football analogy, the two cards overclocked, are simply not in the same class. A purchaser might get lucky and get a non reference 390x way above the standard deviation (say 50% above the TPU test) but that 10.5% OC doesn't get it anywhere near the consistent 30+%improvement we saw from the non reference 980 Tis
Other red herrings ...
-BF3 favors nVidia ... so what, we didn't look at how much better each card was in each game relative to another, we measured how much faster each card was when overclocked. Coding favoring one card or the other is irrelevant.
-Manufacturers always hand pick cards sent in for review ... if we accept that premise, we must accept that both camps are doing the same thing and therefore irrelevant.
-Can't compare different vendor cards as one may have better class of materials on PCB ... both cards were from same manufacturer and same "Gaming Series".
Now if ya wandering why I didn't choose the Fury X, again it was not about the relative performance of the two cards but about how each one overclocks. In this vein, the Fury X would make an unfair comparison as:
a) The Fury X only overclocks 5.1% 108.1 fps overclocked versus 102.9.
b) There is no such thing as a non-reference Fury X
There are threads on overclocking a 4670k and if you are looking for 4.8 Ghz, your definition might be applicable. But look at OC threads on OC oriented sites like overclock.net and you see spreadsheet compilations of member overclocks numbering in the 100s... I don't see anyone with adequate cooling not getting to 4.4 or 4.5 Ghz. So if we are seeing OC's on 4670k's of 13% to 26%, that doesn't mean another CPU doing 4 - 8% would be anywhere in the same ballpark.
After you read 18 reviews on a particular card and every one of them gets between a 26% and 31% over "stock", while another never does better than 4 - 8%, I'd say that's far from a crapshoot. The overall market might be seeing 24 - 33% versus 2 - 10% but that doesn't mean you'll get even odds that that any particular top tier 9xx series wouldn't overclock better than any top tier 3xx series.
By my definition, those are odds I'd take at any casino. That's why they have spreads in football betting. Because it is certainly not a "crapshoot" when a team like Ohio State plays a team like Rutgers... we see that, when tested, OSU is 6-0 and Rutgers is 2-3 and tho any team can lose on any given day, that doesn't make it a "crapshoot". The betting line is not about who will win the game but just by how much.... the two teams are just not in the same 'class of performance". So the "crapshoot" when comparing the two top GFX cards, is not about who wins, just about by how much.
When doing our own or user builds, we base the purchasing decisions on how each custom PCB is built and whether it has proven via published testing that the performance reflects what is expected from the componentry used. The "crapshoot" as you call it, or what mathematicians call the "standard deviation", means that if you have 15 samples test between 25% and 31%, then the variation in the general piopulation might range from 22% to 33%. In no way will the card that OC's only 4-8%, ever ever get up to thew 20% range. The fact is, the R9 series is very aggressively overclocked 'in the box", leaving very little headroom. After all, the 7xxx series were tweaked a bit to make the R9 series and the R9 200 series was tweaked some more to make the R9 300 series. There's simply just no more room to be had. And the new Fury series has less OC headroom than the R9s.
When you look even at the performance of the same card from various vendors, you see trends. For example on the 970s, the Gigabyte G1 wins almost every head to head comparison. Both MSI and Gigabyte routinely break 1500 Mhz boost clock, whereas the competition from other Big 4 vendors fails to break that barrier in just about every review. With our 7xx based builds, the lowest OC experienced was 22%, highest was 28% ... with R9 200 series, the highest was 16 % and that single build stood well above all the others
So rather than a "crapshoot", the data shows an "overwhelming probability of result". Techpowerup recorded a 26% OC (1090 / 863) over reference speed in their OC review of the Asus 780. On the box I am typing from, (twin water cooled 780's in SLI), I managed 1078 (25%). On a subsequent MSI 780 build, we got 1104 MHz (28%).
Every piece of silicon is different, but there is a determinable statistical range which can reasonably be expected. The reference 980 Ti has scored 102.6 fps in BF3 as tested by techpowerup showed. Aftermarket cards hit:
Palit GTX 980 Ti JetStream = 133.1 fps (30% OC)
ASUS GTX 980 Ti STRIX = 131.7 (28% OC)
ZOTAC GTX 980 Ti AMP! = 130.4 (27% OC)
MSI GTX 980 Ti Gaming = 130.5 (27% OC)
Gigabyte GTX 980 Ti G1 Gaming = 134.8 (31% OC)
EVGA GTX 980 Ti SC+ = 126.6 (23% OC)
That is an average of 1.28 .... 1.29 if you eliminate the EVGA card which is logical given that it's not really a non-reference card ... it's a non-reference cooler on a reference PCB.
The standard deviation for the true non-reference cards is just 1.8% .... not exactly what you could call a "crapshoot".
OTOH, the 390x results are quite different. The reference 390x scored 84.3 fps in BF3 as tested by techpowerup. The MSI 390x managed just 7% more. It think that we can expect MSI to put the same effort into making the 390x non-reference card as they did their 980 Ti and yet one managed a 31% OC and one managed just a 7% OC. Given the consistency of test results and standard deviation shown above, the only conclusion that can be drawn from the above that the difference is related to the GPU.
Again like the previous football analogy, the two cards overclocked, are simply not in the same class. A purchaser might get lucky and get a non reference 390x way above the standard deviation (say 50% above the TPU test) but that 10.5% OC doesn't get it anywhere near the consistent 30+%improvement we saw from the non reference 980 Tis
Other red herrings ...
-BF3 favors nVidia ... so what, we didn't look at how much better each card was in each game relative to another, we measured how much faster each card was when overclocked. Coding favoring one card or the other is irrelevant.
-Manufacturers always hand pick cards sent in for review ... if we accept that premise, we must accept that both camps are doing the same thing and therefore irrelevant.
-Can't compare different vendor cards as one may have better class of materials on PCB ... both cards were from same manufacturer and same "Gaming Series".
Now if ya wandering why I didn't choose the Fury X, again it was not about the relative performance of the two cards but about how each one overclocks. In this vein, the Fury X would make an unfair comparison as:
a) The Fury X only overclocks 5.1% 108.1 fps overclocked versus 102.9.
b) There is no such thing as a non-reference Fury X
CptBarbossa
Honorable
- Jan 10, 2014
- 401
- 0
- 10,860
Wow. Apparently I offended someone.
I agree your likelihood of getting a good overclock is better on an Nvidia card than an AMD card.
Having said that, I, as well as most of my friends, don't overclock our GPU's. I just bought an r9 390 that is still at stock. My friend has a 980ti he leaves at stock. I have yet another friend who also has an r9 390 at stock clocks. None of us tends to overclock our GPU's (I do overclock my cpu's quite a bit). In fact, I would say of all of my many gamer friends with custom computers (15-20), I only know of 3 that ever overclock their graphics cards.
Overclocking may be a major factor for YOU, but it is certainly not for everyone.
I wanted the best out-of-the-box experience I could get for my money at the resolution I play it.
You have done your research, which is impressive. I am guessing you would have gotten information from many sources regardless of HOW Toms made their suggestions, making your complaint seam a little unnecessary. I know I checked out a lot of sources when I chose my GPU, so I can totally respect someone who puts that kind of effort into finding what they want.
Perhaps with all of that information you have looked up you could write some articles for Toms. I am definitely interested in that information and I know a lot of people would find it useful. That is just not where Toms chose to focus with these suggestions.
when i started buying cards in 2005 this section really helped. so dear editor if planning on going on hiatus again, please inform us users beforehand. Good job though on your opening defense for the hiatus. I mean as an avid reader the reviews had become too obvious yet times had really changed
It's good to have this article back. I think a quarterly update is sufficient, if that allows you to put more time into the analysis each time. A couple of points:
1. I agree with other comments on the viability of the 390 over 970 and Fury over 980. More credence should be given for these cards.
2. More than once you tout the advantages of the blower cooler. However, for the 980ti, it has been proven in numerous reviews that the stock cooler is not sufficient. It is loud and results in thermal throttling. So while exhausting heat out of the case is nice, at a minimum I would expect the cooling capacity to be sufficient to not require throttling. Aftermarket coolers by Asus, EVGA, Gigabyte, MSI and even Zotac all provide better cooling capacity and a much quieter gaming experience. Yes, you need a case with plenty of airflow capacity, but if I'm running a case that can handle the heat of even an R9-280X being dumped into the case, then it can handle the 980ti with the more power-efficient maxwell design.
1. I agree with other comments on the viability of the 390 over 970 and Fury over 980. More credence should be given for these cards.
2. More than once you tout the advantages of the blower cooler. However, for the 980ti, it has been proven in numerous reviews that the stock cooler is not sufficient. It is loud and results in thermal throttling. So while exhausting heat out of the case is nice, at a minimum I would expect the cooling capacity to be sufficient to not require throttling. Aftermarket coolers by Asus, EVGA, Gigabyte, MSI and even Zotac all provide better cooling capacity and a much quieter gaming experience. Yes, you need a case with plenty of airflow capacity, but if I'm running a case that can handle the heat of even an R9-280X being dumped into the case, then it can handle the 980ti with the more power-efficient maxwell design.
cub_fanatic
Honorable
- Nov 21, 2012
- 1,005
- 1
- 11,960
Newegg has a 750ti 2GB Zotac for $110 w/o any rebates as well as a 2GB PNY for $99 with a rebate. Those prices have been that way for a while, too. So, they do go below $120 quite often. A few months back, I even saw one, a 750ti 2GB MSI I believe, for under $80 with a rebate and huge discount - and it was new not open box or refurbished. The only thing that sucks, though, is that they haven't trimmed vanilla GTX 750 prices. The GTX 750 1 GB should be a $65 card now or less and the 2GB should be around $80. But, on newegg, that $110 750 ti 2GB with no rebate is actually cheaper than some 750 1GB cards. This is completely ridiculous and they will never sell the 750 vanillas with these prices because who would pay more for less performance? If the vanilla 750 1GB was in the $65 range, it would be a hell of a budget card since many reviews have shown that in 720p and 1080p with medium detail, there isn't much difference between a 750ti 2GB and a 750 1GB.
VaporX
Honorable
- Jan 20, 2014
- 21
- 0
- 10,520
Right now the two best values seem to be the 380 and 390. Both cards have solid leads at their price points and deliver great gaming experiences. Not sure why anyone would think the 970 is a better buy than a 390, better performance, better memory system, same price, would seem a no brainer. I guess the competition pays more ads money.
kiniku
Distinguished
- Mar 27, 2009
- 247
- 68
- 18,760
For all the AMD saviors/spazzes trying to split hairs, most the Nvidia products are more efficient, run cooler, SLI works, receive and have received timely driver updates, and have healthy overclock headroom for those that wish to. As a matter of fact the upper end Nvidia products have plenty of factory warrantied overclocked models to choose from. Enough already.
Brian_R170
Honorable
- Jun 24, 2014
- 288
- 2
- 10,785
I was hoping to upgrade my Radeon 6950 with something in the $250 range. Is there really nothing worth mentioning between $190 and $320? Really!?!?
I guess a ~$320 card is what I'll be buying. The GTX 970 is the clear winner for me just for the difference in power consumption. It's 27% lower than my current 6950 47% lower than the competing R9 390.
I guess a ~$320 card is what I'll be buying. The GTX 970 is the clear winner for me just for the difference in power consumption. It's 27% lower than my current 6950 47% lower than the competing R9 390.
I am very disappointed in this article. Usually I agree with a lot of the conclusions, but I can't this time.
The gtx 970 and the r9 390 both retail for around $300 right now. There is no reason a gtx 970 should be recommended above the r9 390 if you are being objective about price to performance.
Heck, the r9 390 should be the recommended card over the gtx 980.
The gtx 970 and the r9 390 both retail for around $300 right now. There is no reason a gtx 970 should be recommended above the r9 390 if you are being objective about price to performance.
Heck, the r9 390 should be the recommended card over the gtx 980.
- Status
TRENDING THREADS
-
Question Using cloning software for a dying HDD which shows only 10% health ?- Started by Mashuumatics
- Replies: 1
-
-
News Windows 11 will reportedly display a watermark if your PC does not support AI requirements- Started by Admin
- Replies: 22
-
Question Recovering an old version of a Notepad file in Windows 10 ?- Started by MC72
- Replies: 1
-
Facebook
Twitter
Instagram