Having the top single and multi-GPUs and retaining a larger market share?
What is your point here?
Having the top single and multi-GPUs and retaining a larger market share?
What is your point here?
Sure the consumers see the benefit. Do you think if the 4870 didn't perform as well as the GTX260 we wouldn't have all benefited from the price war? AMD was able to price their cards where they did because of the design decisions they made for the RV770. Did DDR5 make the 4870 faster than a GTX260 216 with DDR3? No, they are very close in performance. But, the 4870 gave users that level of performance for under $300 vs. $450 Nvidia wanted for a GTX260 192 part at launch. That IS a benefit.
Looks like they were the only portion of AMD that turned a profit. The decision to buy ATI doesn't look like a terrible decision to me... now how much they overpaid, that's a different story.
cbn
Lifer
- Mar 27, 2009
- 12,968
- 221
- 106
I hear what you are saying but if ATI can keep die size down and total costs reduced (relative to Nvidia) future-wise they won't neccessarily need to have the fastest card. They just need to be first to market with the new generation of GPUs that at least significantly "one-ups" the previous generation of top performing cards (from any manufacturer)
Admittedly that GT200 core is a true world beater from a raw performance aspect , but how much does it cost to make considering how large it is?
Being better is one thing but lets say (for the sake of argument) competition can make something 90% as effective for 60% of the cost this is going to impact future manufacturing decisions. (Re: this impacts the ability to "force" price cuts on the competition)
Wreckage
Banned
- Jul 1, 2005
- 5,529
- 0
- 0
I don't care how much it costs to make. Do you? I got mine for around $200 and you can now buy an overclocked core 216 for around $180. Besides if it's financials you are worried about. NVIDIA has something like $2 billion in the bank while AMD could be out of business by the end of the year.
Do you think that if the 4870 was the top GPU they would have sold it for the same price? Nope.
I'm sure they would have charged more than $299. The point was that DDR5 being a design decision very likely did provide a benefit to consumers. Whether you wanted an AMD/ATI card or an Nvidia card, pressure from the 48x0 cards forced Nvidia to lower prices so you could get the card you wanted for less money. That is a benefit to the consumer.
akugami
Diamond Member
- Feb 14, 2005
- 6,210
- 2,552
- 136
The roadmap to smaller manufacturing processes are just that, signposts that say we're going to go here and here and here. One still needs to research the materials and do testing to see how it works out. And just because it is the same nm size manufacturing process doesn't mean you use the same road to get there. Different materials used can lead to different yield percentages, heat, power consumption.
All of the above factors affect the performance and availability of CPU's and GPU's. Costs savings are usually passed down to consumers (except when there is a lack of competition such as the ATI gaffs where nVidia has benefited the last couple years) and as consumers this is one of the ways we benefit but they also help keep the companies who produce the products we buy in business when they are able to save money.
If no one cares about how a card is made why the emphasis on how nVidia's GPU is great because they have the fastest single GPU solution? ATI was beating nVidia on the price/performance ratio and this forced nVidia to drop it's prices.
Even after the price drop ATI cards of roughly equal performance, specials and occasional large rebates notwithstanding, could be found quite often for less than the competing nVidia card.
I'd say that consumers benefited a huge amount because ATI was cleaning up the charts on the price/performance standpoint forcing nVidia to drop it's prices to stay competitive. How is that zero benefit to the end user?
Originally posted by: OCguy
Where are the faster and cooler DDR5 cards?
Imagine how hot the ATI cards would run without the process shrink?
As for DDR5, this helped reduce the overall costs associated with ATI card design. You can't say that after all it's said and done the smaller ATI GPU's along with simpler board design didn't help it from a cost perspective. Sure, heat wise it might be similar to nVidia cards and past ATI cards as far as heat output during full load but the cost savings has benefited nVidia buyers when ATI forced nVidia to drop it's prices to stay competitive.
True but what's done is done and it is the CPU business that's dragging things down (the graphics division was profitable in at least one quarter in the last little while IIRC). Regardless of whether they would have had the extra cash, they need to change things as it would have gotten to the point they are at now at sometime if they keep going as they are...it just happened quicker due to several circumstances (buying ATI being one of them).
Wreckage
Banned
- Jul 1, 2005
- 5,529
- 0
- 0
Well either way the economy has been $hitty to everyone. I look forward to the day when we can recommend people buy $1000 video cards because everyone is doing well enough to not consider that a big deal.
As for the original topic. Until there is a hard launch of a DirectX 11 part from either company we are basically talking about a mythological card that only exists in the minds of dreamers.
Relatively speaking AMD did well then too. Their losses were less than anticipated. Edit: I just thought about it and figured you were talking about Intel vs. AMD, not Intel vs. expected earnings. Sorry :
As for the topic. This could be similar to the 9700 Pro. Didn't it release prior to DX9?
cbn
Lifer
- Mar 27, 2009
- 12,968
- 221
- 106
http://forums.anandtech.com/me...=2296969&enterthread=y
In light of GT300 (more than double the processing power via 512 processors....up from GT200's 240 processors) what do you think ATI will be doing with HD58xx?
Will they be able to fit 2000 stream processors on their current die size? Or will they use something more conservative like 1600 stream processors?
Maybe (in this economy) a 1600 stream processor design for ATI is not a bad idea provided they can release the new part before Nvidia.
Something just tells me the market for the ultra high-end GPU may be a little lower than what is was previously and making GPUs with purposely deactivated cores (to fill in lower price points) may not work as well if the starting core is too expensive or big?
In light of GT300 (more than double the processing power via 512 processors....up from GT200's 240 processors) what do you think ATI will be doing with HD58xx?
Will they be able to fit 2000 stream processors on their current die size? Or will they use something more conservative like 1600 stream processors?
Maybe (in this economy) a 1600 stream processor design for ATI is not a bad idea provided they can release the new part before Nvidia.
Something just tells me the market for the ultra high-end GPU may be a little lower than what is was previously and making GPUs with purposely deactivated cores (to fill in lower price points) may not work as well if the starting core is too expensive or big?
it's hard to say. going from 3870 to 4870 was huge, and AMD won this round for the most part. Their cards are very cost effective. Going from G92 to G200 was a disappointment, and you've got tons of kids with G80/G92 cards who will probably upgrade this time around. It's AMD's job to make sure GT300 isn't a $600+ card at launch. if anything, the market is growing, but people can't upgrade every season if the gains are going to be less than 20%.
Yeah ATI's problems the last few years seemed to revolve around their memory bus. First, with the X1800 cards, they introduced the 512-bit ring bus, but were also 6 months late with problems in the X1800 card.
They recovered nicely with the X1900 series, but then had massive problems bringing out the 2900XT (still the only true 512-bit card to date).
"Brute Force" architecture hasn't truly worked for ATI since the 9700 Pro, meanwhile, Nvidia has hit home run over home run (6800 series, 8800 series, GT200 series).
ATI seems to operate best being "lean and mean" - the 38xx and 48xx series were successes because they built them to be performance per watt champs and cheap to make.
Hehe, that post is quite comical. You realize that the x800 Phantom Edition was clocked at 500MHZ right? Check out the clock rate on that engineering sample review you linked. Limiting ourselves to actual released parts, nV did come out ahead that generation.
It overwhelmingly obliterated the x1950xtx- the review you linked even shows that. Yes, some benches the x1950 pulled ahead, just as the 192 core 260 sometimes bested the 4870x2, doesn't mean that the 4870x2 is very accurately considered a MUCH faster board. As far as why the comparison was made in the first place, someone made the assinine assertion that none of the 7 series GeForce products were faster then the x19 series, this very clearly wasn't the case.
Versus the PE which didn't exist either? Sorry, if we are comparing imaginary cards then green teams should be compared to red teams. If we eliminate the imaginary cards, then nV wins the majority of benches, even in the review you linked.
You could SLI 7900GX2s also.
So why not compare that with x1900 crossfire? You do know that looked even worse on nVidia's side right? Because of that original look at dual sammiches, I'm truly astonished today when I see tri- and quad- setups even posting any positive percentage benefit over dual-cards.
The X800XT PE was definitely produced and sold to the public. In fact, my daughter's computer is running one right now.
cbn
Lifer
- Mar 27, 2009
- 12,968
- 221
- 106
x1900 Crossfire was better than 7950GX2. but at that time ATI didn't have a dual GPU card of their own.
Speaking of dual GPU Video cards when do you think we will see Tri-SLI or Tri-Fire in a single PCI-E slot?
If this ever happened hopefully someone figures how to get all the memory contributing instead of this "mirroring" effect (or whatever it is called). This mirroring effect with the memory is just a big waste of chips.
Ok, what is supposed to change my view? I guess if you drop the settings way down, then quad sli didn't look so hot. Crank everything up, and quad sli actually rather handily bested either normal SLI or Crossfire for that generation.
I suppose that is true, to be completely fair ATi did sell some of the x800xt PEs based on various reports. But availability seemed to be comparable however.
When chips are small enough and cool enough with reasonable power requirements would be my guess. ATi seems to be headed in that general directon.
That would be extremely difficult at best. Likely the best way to approach that would be to have three seperate chips that do differing things on a shared high speed bus connected to a common memory pool. I say that would likely be the best in as much as that would be much easier to engineer then trying to get our current multi chip approach to not require redundant memory. Not saying it isn't possible, and I'm sure both nV and ATi have people working on that, just if it does come down the pipe it will likely be extremely complex(and I would assume who ever gets it done first will patent the hell out of it making it nigh impossible for the others to follow).
cbn
Lifer
- Mar 27, 2009
- 12,968
- 221
- 106
I'm thinking this might even save on development costs in some scenarios.
The current trend for Nvidia right now appears to design a "mega core" video card and then either deactivate part of it or just use the old cards for the midrange and lower range. But then for the lower range how easy will it be to downsize G92 to 40nm for this purpose?
At some point they will need to design something new for these lower cost SKUs and doing this with an entirely new creation costs money.
No, the "Phantom Edition" was clocked at 520mhz. I used to have a regular x800xt, and there was nothing phantom about it, not to mention it easily ran at 520+mhz.
LMAO, by that measurement the x800xt also "overwhelmingly obliterated" the 6800U.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 403
-
Facebook
Twitter