-
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.
Release Dates - Radeon HD 7000 Series
- Thread starter Sparx
- Start date
+1, this is the one i'm looking out for.
There is just something about XT that makes you want it more
Still remember my 9600XT ahhh Far Cry and BF2
Soldato
- Joined
- 22 Nov 2007
- Posts
- 3,763
- Location
- Ayr, Scotland
No mate the 7850\70 are mid-range ie 6850\70 therefore we should be talking £130-£160. Remember the 6970\50 are the current high end, not counting 6990 as thats extreme, ie £220-£350, so I'm hoping for a 7870 for £180 including the usual price gouging for new cards or 2 X 7850 for £280 including price gouging. These prices should fall back a bit as time passes and there's more availability.
I would take a guess that it would be bigger than the 69xx cards, but not by that much, maybe 15% tops. From what I've seen a 6970 is only giving 7-10% more then a 6950 depending on OC. Just hope the 7950's are priced the same as the 6950's were. And I don't think AMD will be making the dual bios decision again, that must have cost them a fair few 6970 sales.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Impossible to know, the 5850 was awesome in that the clock speeds were so stunted, it was one of the reasons the price difference between that and the 5870 was so large, if it had only 10% less shaders but was clocked 50-75Mhz higher then the gap could have been smaller.
Personally I hope for a smallish shader count difference and a large drop in clock speed(but hopefully not too pathetic a cooler to go along with the lower clock), that will be best for price difference and easily regainable performance through overclocking.
But who knows, if it turns out 7950's unlock again, overclock great, have voltage control and the same heatsink........ woo, it could have 10% less shaders, not overclock, have a hefty drop in clock speed and the AIB's cheap out on a much less capable heatsink, and no voltage control making overclocking up to 7970 levels and beyond almost impossible.
From what I recall though, 4870 800 shaders, 4850, 720 shaders, 5870 1600 shaders, 5850 1440, 6970 1536 shaders, 6950 1408 shaders. So its generally 10% of the closest they can get to it. It will come down to yields and where salvaged parts get the best number of cores off the wafer. AFAIK there is 64 shaders per CU.
The 6970/6950 was 2 CU's disabled which was just under 10% of the shaders disabled. Now at circa 2000 shaders, 2 clusters would be noticeably less than 10% shaders(so we can hope for that and a bigger clock drop) but they might go for 4 clusters for yields and because 2 clusters is only just over 6% drop in shaders and generally speaking symmetry in numbers is pretty normal.
So 6.5% less shaders and a big clock drop is what we can hope for, 13-14% less shaders and a higher clock might be what we get.
That or this gen we could see say a 7970 at £300, a 7950 at £250(significantly higher, with a much smaller drop in performance, 6% less shaders 5% lower clock), and a 7930 not long after with the full 13% shader drop, 10% lower clocks, and in at that £200-215 price point.
7950s in Crossfire.. Mmh.
Thats what I'm waiting for aless the 7990 comes out before or not long after.
Soldato
- Joined
- 18 May 2010
- Posts
- 23,716
- Location
- London
Report here has some performance guesstimates as well as some potential bad news....
You know the funny part of all this ? Suppliers are trying to empty their stock of old cards and with all these delays to the new cards they will end up with no high end cards in stock and customers heading off to other suppliers that used their common sense. Also that performance guesstimate is another AMD classic leaking some numbers that mean nothing without adding the settings and hardware used "single-GPU cards should be able to get more than P45,000 marks in 3DMark Vantage" ..sure it did.. and on what settings 640 x 480 ? that guesstimate is useless information without knowing the full settings and hardware (CPU/MOBO/RAM) they used to run the test, as useful as "AMD Bulldozer breaks own world record, overclocked to 8.46GHz" yes it hit a record overclock speeds but it also ended up a record FAIL for the company on many accounts, also they should be added into the record books for the CPU that wastes the most electricity.. Anyone seeing the same trend AMD is doing ? I smell another Bulldozer with the GPU section... grrr.. I also feel Nvidia is going to let us down too with their GPU line up this time round.. All the new changes to their cards architecture and the manufacturing process to make them is equaling BETA cards to me will be available to consumers and we as enthusiasts are going to be their BETA testers. I hate to see the mess with their drivers at AMD/ATI if they can't even make the hardware. Their hardware people in my book are top notch and the only thing that has started letting AMD/ATI down in recent years is their driver team.
AMD Radeon HD 7900 Tahiti GPUs to Arrive in January 2012 – Report
http://news.softpedia.com/news/AMD-...to-Arrive-in-January-2012-Report-233382.shtml
Radeon HD 7900 series in January
http://www.fudzilla.com/graphics/item/24778-radeon-hd-7900-series-in-january
*YAWN*
"In addition, both Radeon HD 7950 and Radeon HD 7970 reference designs will come with an improved cooling system that replaces the traditional vapor chamber technology used for high-end parts with a liquid chamber."
Liquid chamber LOL... This thing is going to run Bulldozer heat and electricity it seems too. That is if any of this info is real.. it is softpedia and fudzilla reporting all this.. or should I say softpedia copying other sites so called news as always.
The only thing I took from the Fudzilla report as being 100% true is this part..
"Hopeful 2012 will be a bit more interesting in the graphics world, as 2011 will probably go down in history as one of the most uneventful years since the nineties."
AMD Radeon HD 7900 Tahiti GPUs to Arrive in January 2012 – Report
http://news.softpedia.com/news/AMD-...to-Arrive-in-January-2012-Report-233382.shtml
Radeon HD 7900 series in January
http://www.fudzilla.com/graphics/item/24778-radeon-hd-7900-series-in-january
*YAWN*
"In addition, both Radeon HD 7950 and Radeon HD 7970 reference designs will come with an improved cooling system that replaces the traditional vapor chamber technology used for high-end parts with a liquid chamber."
Liquid chamber LOL... This thing is going to run Bulldozer heat and electricity it seems too. That is if any of this info is real.. it is softpedia and fudzilla reporting all this.. or should I say softpedia copying other sites so called news as always.
The only thing I took from the Fudzilla report as being 100% true is this part..
"Hopeful 2012 will be a bit more interesting in the graphics world, as 2011 will probably go down in history as one of the most uneventful years since the nineties."
Last edited:
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Jesus H christ people, firstly Softpedia are actually worse than Fud, they link a XTREMESYS post as their source, in fact they don't, they STATE it as a source, without even linking the post, at least Fud and some other idiots do that. Softpedia are wrong, on everything.
Not least that, Tahiti is both unlikely to be a very large die size and not a single leak so far has suggested 384bit bus afaik which makes that news story even more ridiculous.
They go on to claim low yields at 28nm, 28nm doesn't have low YIELDS it has low CAPACITY, which are entirely different things. A bunch of news sites, Softpedia included keep stating that low capacity is an indication of problems, like its new and never before happened. Every single new process, EVER, has started off with low capacity.
You know how you can't actually get to 100mph, without first starting at 0mph, then going through 10, and 20, and 30mph, etc, etc. Yes, production starts off at nothing and increases from there, this is how every process has ever gone.
Also Purgatory, nothing you said has any bearing on anything at all, liquid cooling, is simply a little liquid in a vapour chamber type device, instead of the vapour chamber, because they leak and fail at a semi high rate, its not "water cooling", its nothing like watercooling. Likewise Sandybridge-e uses more power than Bulldozer, has only 10% more transistors yet is 35% bigger or so, and its their last new chip on 32nm, designed for 32nm from the ground up, Bulldozer is the FIRST chip designed for 32nm, on the first gen of a new architecture(this is always, always the least efficient, how good were the first P4's, or the 754 Ath 64 vs even a 939 one, let alone Phenoms and Thubans).
At least you managed to realise Softpedia are rubbish, seriously if any site should be banned from being linked, Charlie gets a lot of bad press, he does like an Nvidia rant, so what, he's accurate a LOT of the time and breaks more REAL exclusives than any other tech reporter around. Softpedia are whores, they post forum posts as confirmed info, and the make Fudzilla look like an almost professional site.
The funniest thing is when Softpedia or Fudzilla pat themselves on the back for confirming a news story, that is normally still wrong, that is just a rumour that every other site has been running for over a month anyway but they decide to take as fresh news exclusive to them.
Seriously, they even make the Daily Mail look professional.
Not least that, Tahiti is both unlikely to be a very large die size and not a single leak so far has suggested 384bit bus afaik which makes that news story even more ridiculous.
They go on to claim low yields at 28nm, 28nm doesn't have low YIELDS it has low CAPACITY, which are entirely different things. A bunch of news sites, Softpedia included keep stating that low capacity is an indication of problems, like its new and never before happened. Every single new process, EVER, has started off with low capacity.
You know how you can't actually get to 100mph, without first starting at 0mph, then going through 10, and 20, and 30mph, etc, etc. Yes, production starts off at nothing and increases from there, this is how every process has ever gone.
Also Purgatory, nothing you said has any bearing on anything at all, liquid cooling, is simply a little liquid in a vapour chamber type device, instead of the vapour chamber, because they leak and fail at a semi high rate, its not "water cooling", its nothing like watercooling. Likewise Sandybridge-e uses more power than Bulldozer, has only 10% more transistors yet is 35% bigger or so, and its their last new chip on 32nm, designed for 32nm from the ground up, Bulldozer is the FIRST chip designed for 32nm, on the first gen of a new architecture(this is always, always the least efficient, how good were the first P4's, or the 754 Ath 64 vs even a 939 one, let alone Phenoms and Thubans).
At least you managed to realise Softpedia are rubbish, seriously if any site should be banned from being linked, Charlie gets a lot of bad press, he does like an Nvidia rant, so what, he's accurate a LOT of the time and breaks more REAL exclusives than any other tech reporter around. Softpedia are whores, they post forum posts as confirmed info, and the make Fudzilla look like an almost professional site.
The funniest thing is when Softpedia or Fudzilla pat themselves on the back for confirming a news story, that is normally still wrong, that is just a rumour that every other site has been running for over a month anyway but they decide to take as fresh news exclusive to them.
Seriously, they even make the Daily Mail look professional.
drunkenmaster I was being sarcastic in most of what I said above mate. Softpedia and Fudzilla are well how do I put this polietly... Morons and make news out of nothing or the very obvious. Reality is we don't know till they land in the stores or in the hands of people that will review them. All these guesstimates of so called benchmark tool results is nothing more then AMD's marketing team leaking to people like them and the reality is some will read it and think wow it's real and people in the know and that have followed hardware development know it's just pure smoke and mirrors. Can't wait for them to come out from both teams, but i'm looking more forward to what Kepler will bring to the table and what the new memory that AMD is using will bring too.
The reporting of both sites is a joke and they need to learn how to spell and correct basic grammar mistakes. (Ninja Edit)
Regarding the liquid cooling, I do see these cards running hot to be honest just like how SB-E is showing on the 32nm and IB-E on the 22nm (when it appears) will still run toasty compared to SB 2500k,2600k and 2700k standards I bet, as we see reducing the node size in the manufacturing process or architecture changes does not mean your next generation will run cooler or more efficiently. SB-E uses more power then the last generation intel 6 core.. yes it's faster and so it should be, but very power hungry and not the promises we were told about reduced power and heat with the manufacturing process or architecture changes (YES I know both generations use 32nm). Of course it reduces it to a point but depends at the end on the devices architecture too and how it processes the data. Honestly we should all just wait and see, node changes and new architecture means nothing unless it shows it can do things better then the last generation. Bulldozer is a great example of a *FAIL* poor performance, very unefficient and produce lots of heat and SB-E is a great chip performance wise but also in my book a FAIL because again they are not efficient and produce a lot of heat and they removed what I thought were key new features to SB, like quicksync and SRT. Also SB-E is an 8 core 16 threaded cpu, that has had 2 of its cores disabled and some cache disabled too, can you imagine what the real wattage would have been, they kept it at 130w and of course the ones with fully functioning 8 cores will end up as Xeons for the server market and the extra cache enabled. SB-E reviews today showing power use at idle and full load made me realise why we still don't have 8 core cpus from intel yet in our homes, the problem is power use and heat. In a server environment that can all be controlled. I don't even think Ivy Bridge-E will have more then 6 cores active again and again the Ivy Bridge-E with fully working 8 cores and cache will end up as Xeons.
I was waiting for SB-E and after reading some previews I decided this was not a chip for me and not a chip I want to have to sell in the future to fund a new build with. It is over priced and the performance jump from the previous generation is not that huge, better to invest in a X58 setup at that point I thought but decided to forget that too and spend more money on a SB 2600k and buy a better quality motherboard, more RAM while the prices are so cheap and a new graphcis card.
I honestly don't regret it even after reading and watching some of the reviews today about SB-E. It still is not for me sorry to say. If I was rendering or running a massive database that needs all that RAM that X79 has access too then yes, but for a power users home pc it is overkill to the max is SB-E and will not show any real world advantages in that situation.
So you see why I don't listen to all this blaa blaa about the new GPUS ? because reality is they can turn out either amazing or a total flop... my bets are so far on they will be a flop and nothing we have not seen before with again another minor speed bump from both camps while they get use to using the new manufacturing process or architecture changes, next generation I feel we will see what they can really do.
The reporting of both sites is a joke and they need to learn how to spell and correct basic grammar mistakes. (Ninja Edit)
Regarding the liquid cooling, I do see these cards running hot to be honest just like how SB-E is showing on the 32nm and IB-E on the 22nm (when it appears) will still run toasty compared to SB 2500k,2600k and 2700k standards I bet, as we see reducing the node size in the manufacturing process or architecture changes does not mean your next generation will run cooler or more efficiently. SB-E uses more power then the last generation intel 6 core.. yes it's faster and so it should be, but very power hungry and not the promises we were told about reduced power and heat with the manufacturing process or architecture changes (YES I know both generations use 32nm). Of course it reduces it to a point but depends at the end on the devices architecture too and how it processes the data. Honestly we should all just wait and see, node changes and new architecture means nothing unless it shows it can do things better then the last generation. Bulldozer is a great example of a *FAIL* poor performance, very unefficient and produce lots of heat and SB-E is a great chip performance wise but also in my book a FAIL because again they are not efficient and produce a lot of heat and they removed what I thought were key new features to SB, like quicksync and SRT. Also SB-E is an 8 core 16 threaded cpu, that has had 2 of its cores disabled and some cache disabled too, can you imagine what the real wattage would have been, they kept it at 130w and of course the ones with fully functioning 8 cores will end up as Xeons for the server market and the extra cache enabled. SB-E reviews today showing power use at idle and full load made me realise why we still don't have 8 core cpus from intel yet in our homes, the problem is power use and heat. In a server environment that can all be controlled. I don't even think Ivy Bridge-E will have more then 6 cores active again and again the Ivy Bridge-E with fully working 8 cores and cache will end up as Xeons.
I was waiting for SB-E and after reading some previews I decided this was not a chip for me and not a chip I want to have to sell in the future to fund a new build with. It is over priced and the performance jump from the previous generation is not that huge, better to invest in a X58 setup at that point I thought but decided to forget that too and spend more money on a SB 2600k and buy a better quality motherboard, more RAM while the prices are so cheap and a new graphcis card.
I honestly don't regret it even after reading and watching some of the reviews today about SB-E. It still is not for me sorry to say. If I was rendering or running a massive database that needs all that RAM that X79 has access too then yes, but for a power users home pc it is overkill to the max is SB-E and will not show any real world advantages in that situation.
So you see why I don't listen to all this blaa blaa about the new GPUS ? because reality is they can turn out either amazing or a total flop... my bets are so far on they will be a flop and nothing we have not seen before with again another minor speed bump from both camps while they get use to using the new manufacturing process or architecture changes, next generation I feel we will see what they can really do.
Last edited: