Caporegime
They prioritised R and D for CPUs over GPUs,so sadly I can't see them make an all out assault against Nvidia anytime soon!
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Dark days, AMD share price at lowest ever.
- Thread starter bru
- Start date
They prioritised R and D for CPUs over GPUs,so sadly I can't see them make an all out assault against Nvidia anytime soon!
Man of Honour
- Joined
- 21 Nov 2004
- Posts
- 45,962
Soldato
Look mate AMD still competes on all segments but not on a halo product aka GTX1080Ti, and AMD hasn't since 2015. Last time AMD beat the Nvidia was the R9 290X vs GTX780Ti true.
But I do not remember anyone been so bleak about Nvidia when AMD was beating the Nvidia top offerings of GTX480, GTX580, GTX680, GTX780 (Titan 1) and GTX780Ti (Titan 2). That was almost 6 years on the trot.
And NVidia had nothing to compete with the lower market also. AMD has.
Caporegime
Even when Nvidia had crap products like the FX,which Gabe Newell himself didn't recommend and ran such a mahoosive title like HL2 poorly,they sold more cards than ATI/AMD - even Nvidia mocked the cooler on the FX5800.
So looking at what AMD has planned for the future,they probably are going to push higher end GPUs to things like AI,compute,etc only(7NM Vega for example) where they can make decent margins and sell it with their CPUs as a complete package,stay with consoles where they have more leeway with devs and have something in the midrange(Navi) on desktop as an offshoot of the consoles,since R and D can be reused.
I personally can't see them bothering with the high end now - look at the R9 290 series,people still bought more GTX780 series cards over them,even though Hawaii has been one of the best GPUs AMD/ATI ever created. Six to nine months delay with Fermi and Nvidia still sold more cards than AMD/ATI did during that period. AMD could literally come out with a GPU which beats a GTX1080TI by 30% right now for £500 and consumes less power,and people will still wait for the Nvidia competitor,and say thanks AMD for making their Nvidia card cheaper.
Personally I would rather they concentrate on less areas,and not cover as many markets,since they will probably do better. They have limited R and D spend,so better to concentrate it areas where they can make more money on.
Also,the fact AMD has diverted more of their R and D to CPUs makes much more sense - people are more likely to buy an AMD CPU in the first place than their GPUs,hence why Ryzen has already had a massive effect in such a short time.
I mean,look at what Intel said in its earning call today:
https://www.cnbc.com/2018/07/26/intel-earnings-q2-2018.html
Holiday season means the period from Thanksgiving to Christmas and so on. That is end of November 2019. It makes more sense for AMD to push as much effort into getting Epyc,Threadripper and Ryzen out in quantity on 7nm before Intel can do so on 10NM,as that will most certainly help them more than even their GPUs.
Soldato
R9 290X was a beast of a card that has improved over the years. Still today outclasses the mid range cards completely, including the 9 and 10 series.
GTX780, Titan, 780Ti, Titan X disappeared into oblivion getting beaten by the RX460! last time I checked.
Ha. Remembered the power consumption arguments we had with some people here that period. To pay off the GTX780Ti & Titan X costs by the energy savings over the R9 290X, had to run the 780Ti for 25 years and the Titan X for 40 years. Already those cards were obsolete in just 2.
As for 7nm and Intel 10nm. Intel CEO said just two months ago, 12-18 months.
So yes that is Nov 2019 is inline with 18 months from May.
However that means Zen 3 will be in just couple of months time away, with Zen 2 (what ever it is 12/16 core or 5Ghz 8 core) already 6 months in the market.
Which will make the 10nm by that time 3 1/2 years late or there about.
Only because reviewers persist in using incorrect (often GPU Boost 1 testing methodology when second gen cards don't act like that) testing methods for testing them for some reason - plenty of videos on YT that show those cards perform much better than you see in the lineups of older cards in newer reviews. Power consumption isn't a big issue when you have the leading edge performance to back it up - which those Kepler cards enjoyed for nearly 2 years.
As I posted back with my 780 it was running a hair off the RX480 in games like Doom 2016 (unless you enabled Vulkan) despite reviewers having it languishing back with the 280, etc.
Man of Honour
If you look closer enough they still have !
Volta and Pascal are less efficient than Maxwell in SP gaming.
It is only the higher number of SP cores and clockspeeds that keep Volta and Pascal ahead of Maxwell.
In the gaming area NVidia have not made any efficiency improvements since Maxwell.
Soldato
panos said:
Back when the 970 and 290x were head to head, power consumption was only really a factor to consider when choosing between them. That said, it was still (imo) an important factor, because regardless of running costs, an extra 100w+ of heat is something you need to plan for and of course less heat means less cooling required. I remember well AMDMatt saying his 290x's were pulling a hilarious 300w a card flat out a which was waay beyound anything i could do with my 970. I mean my entire PC with the 970 in it pulled that from the wall. So yes, power consumption is a thing that should be considered, but not because of the running costs. Not for gaming anyway.
Last edited:
Maybe because when those cards came out they beat the competition (or atleast matched it) and when AMD did come out with a card beating them it was usually long enough into the life cycle of the card they'd already sold a lot of volume and new stuff was on the horizon - GTX480 generally competed with the 5870 comparing favourably in most cases with around 10-15% lead over a broad range of games. GTX680 wasn't beaten until AMD re-released the 7970 in GHz guise which produced a small win at the time though matured well in the long run. 780/780ti competed well with a reasonable lead over the 200 series cards for the first year and well into the second year before the R9 cards started overhauling them. On release my 780GHz was competing favourably even with AMD's top overclocked 290X models https://www.bit-tech.net/reviews/tech/graphics/gigabyte-geforce-gtx-780-ghz-edition-review/5/ it was only quite a few months later they clawed back the lead with newer drivers.
AMD has been struggling along in the mid-range GPU wise for quite awhile now with no answer to nVidia's higher end cards and even mid-range is generous - the GTX1070 and 1080 aren't exactly particularly high end cards other than nVidia has artificially held back the performance segments to make it seem like they are :s the specs of the 1080ti versus what the node is capable of would put it in the GTX470, GTX260, etc. spot in any other generation other than the latest madness nVidia seem to be getting away with.
Completely different set of circumstances.
If you look closer enough they still have !
Volta and Pascal are less efficient than Maxwell in SP gaming.
It is only the higher number of SP cores and clockspeeds that keep Volta and Pascal ahead of Maxwell.
In the gaming area NVidia have not made any efficiency improvements since Maxwell.
Rather astonishing isn't it; especially if you can overclock the heck out of a Titan X or 980 Ti for some lovely performance gains.
Sadly, that's what happens when there's little to no threatening competition. Same with Intel; as nice as the upcoming 9900K looks; it's still just improved Skylake, with more frequency and small upgrades.
Now HWInfo has added Volta cards to their latest build; and it might be we have Volta scaled down; instead of big leap from NVIDIA.
AMD increasing their R&D budget by 25% is a start; but I think it'll take 2-3 years and a new GPU architecture design before they're nipping at NVIDIA's top-end again.
As much as many people think GCN is fine; which it is; it's gotten long in the tooth. Even in Compute where it use to lead the way it's now falling far behind. AMD need a new architecture.
What they need is another 5870 vs GTX 480 scenario at best in the coming year; although a 9700Pro vs FX 5800 would be preferred
What they need is another 5870 vs GTX 480 scenario at best in the coming year; although a 9700Pro vs FX 5800 would be preferred
If nVidia try to fleece customers with another incremental release sticking with 12FF and AMD actually put out a stripped and rebuilt Vega on a successful (i.e. probably skip GF) 7nm node you might yet see it.
Ralthough a 9700Pro vs FX 5800 would be preferred
I hope not because AMD actually have to increase their market share not Nvidia. People keep mentioning the 9700 pro as some kind of golden era for ATI. It wasn't. ATI had the best cards on the market without a doubt but still Nvidia sold more cards and increased market share during the time of 9700/9800.
Soldato
A: at best arguably faster in some cases, the 480 was faster but was hotter and more hungry and pricey (clocked like a beast too), the 680 was faster on release and consumed less power and was cheaper than the 7970 , the 780/ti traded blows with the 290/290x/390/390x (all the same chip)
B: They where never missing a product from the high end like AMD has missing now
Look at this way hyperoptic POV. if after Maxwell, Vega released first with a top to bottom product stack and they where fab, a month later Nvidia only released a GTX1060 which competed with the RX480 but consumed more power. leaving 56 and 64 by themselfs.
Almost 18 months later the 1070/1080/1080ti come out with the Ti only competing with a 56 while using a bigger hotter more hungry chip leaving a high end 64 with a large step up in performance all by it self. You and I would be saying Nvidia had issues.
TLDR: you be saying different if the tables where turned.
the current AMD situation reminds me of the FX5800 Ultra or the 2900 pro, hot, slow and late leaving a massive hole on there high ends at the time against the competing products.
You also underestimate how much of a impact having a halo product that faster than the competition has on the other sectors eps in the eyes of job public
Last edited:
It was a golden time though, than 9000 series propelled ATI up, and the x800 series put them ahead. They were competing really well until the 2900 line flopped.
I wouldn't consider that brief blip of less than a quarter a golden era though. Normal service was resuming until Nvidia released one the worst cards ever. My point was, that it wasn't really anything ATI did, it was Nvidia making a major mistake. A little over a year after that, ATI started downward and it's been that way ever since.
Soldato
Last time I checked my RX Vega 64 beats a 1080 or lands just under it depending on title. If that isn't completing with the competition I don't know what is.
Yeah sure you can sit here and rip apart what is and isn't wrong with AMD/nvidia GPUs but the fact remains if they pushing out the frame rates that is all that matters.
Least for me anyway, said it time and time again a Vega 64 at 100+ fps vs a 1080Ti at 150fps I doubt many would tell the difference between them.
The trade off becomes smaller and smaller.
Am glad to see Amd now making money it has shut up all the Doom sayers lol
Yeah sure you can sit here and rip apart what is and isn't wrong with AMD/nvidia GPUs but the fact remains if they pushing out the frame rates that is all that matters.
Least for me anyway, said it time and time again a Vega 64 at 100+ fps vs a 1080Ti at 150fps I doubt many would tell the difference between them.
The trade off becomes smaller and smaller.
Am glad to see Amd now making money it has shut up all the Doom sayers lol
Last edited:
Fine if you are getting those kind of framerates - push up the resolution and settings a bit to where they are both starting to drop towards 60 FPS and the difference in performance becomes much more noticeable. There are some games the Vega cards do very well even able to get within sight of the 1080ti but there is no shortage of games where the 1080ti is in a league of its own especially at 4K type resolutions.
Crossfire/SLI are about as good as it gets when it comes to that approach - even chiplets won't help there if they have to use techniques like AFR (the AMD guy was correct in that respect) - you can't just sew together GPU cores using IF in the style of CPUs and get the same kind of results. (EDIT: Unless developers start building games around DX12 explicit multi-adaptor from the ground up).
The real trick with MCM designs is when you can implement an entire logical GPU on substrate without the restrictions of a monolithic package, possibly using multiple headless processing packages and multiple command processors and arbitrarily scale up/down any one area.
Last edited: