ART of War, nV vs. ATi your thoughts
- Thread starter Razor1
- Start date
Xbot360 said:
Has the 7900's price stabilized at $519? You're jumping the gun a bit. Nvidia would have to be getting some phenomenally horrible yields to not be able to undercut ATi in the high-end segment.
Define "sitting pretty". Things are not as one dimensional as you seem to interpret them to be.
Razor1 said:
You make your own luck as well. Who knows what ATi could have done differently to avoid or intercept this problem in advance (this was part of Nvidia's downfall with NV30). Or even, what they could have done to identify the source of the problem more quickly once the symptoms were evident.
IIRC, this problem arose early enough but it took them a while to isolate and correct the exact source of the issue. Analysis, problem solving and troubleshooting are skills too.
Jawed said:
I'd have thought that it would be within the remit of the 3rd party company who produced the libraries to test them thoroughly. Can ATI be blamed for expecting standard tools/assets to work as advertised? Similarly, didn't NVidia blame TSMC for their problems with Low-K and NV30?
Oddly enough, if a competitor were to now use the same (corrected) libraries they would actually benefit from ATI's problems - a double whammy!
trinibwoy said:
As when comparing any prices, it's all time-sensitive. What's the best value high-end purchase if you need to buy a card this week? Easy: 1900XT. In a couple of months, who knows?
It's much the same argument people seem to have after the release of every new card, comparing early prices to those of cards which have been available for months. We'll be hearing exactly the same arguments later this year over G80 and R600!
Nvidia did have some bad luck with the NV30 but they also had lots of poor planning with it as well.
ATi's excuse that a software glitch screwed up their production process for 6 months is either a pathetic excuse in an attempt to shift blame, or scary they are that incompetent to figure it out in 6 months.
Either way what happened with the R520 and 7800 isnt what I would consider totally lucky. Nvidia planned well with the 6800 and 7800 series and have been hitting every goal since the NV30 debacle. At some point it goes from luck to great planning.
This latest incarnation was probably the most genius yet. They didnt win the speed title, and in some areas lost. But they have a cooler, less power consuming, and much cheaper to manufacture chip that competes. We are talking about a chip with half the die size competing with the competition. That isnt luck, that is great planning. Realizing they have an opportunity to simply hang while reducing production costs and increasing yields, they took it. Why pour money into the chip when you can do the above and focus on the next generation? While ATI fumbles around with a hot, power consumption hungry 48 pixel processor that apparently sits idle most of the time. Nvidia continues on its way and making nice returns for their owners.
While enthusiasts may not be "that" impressed, from a business perspective if you can compete with a product that costs half of your competition it gives you a lot of flexbility in what you want to do.
The OEM market is probably very interested in a lower power, cooler running chip. I am curious when Nvidia will bring out a 7900 variant for their laptops.
My take on the whole war is the R300 was a great hit, slapped Nvidia in the nutz, and got Nvidia back on track. ATi is one of those companies that I dont consider very well run. Before the R300 they werent doing anything but producing low perfoming, bug ridden hardware. After the R300 it is like they cant grasp their own market. When they introduce products in the channel they seem to focus on that speed crown and slowly integrate down the price points. I mean, have they finally killed off the 7500 they have relabeled to 9000,9100,9200? And how long did it take for them to finally get a top down SM3 product line?
The last 24 months havent been very impressive as the X800s either took months to show up or never did and the X1800 showed up late, hot, not overly fast compared to the competition, and was a short lived product cycle ala NV30. The X1900XT looks like a promising chip but I question if it isnt a generation ahead of itself with all that pixel processing power yet barely beats a competing Nvidia part that is almost half its size. And lets not delve into the idiotic design decisions of the X800 crossfire boards, which btw can you even purchase?
ATi's excuse that a software glitch screwed up their production process for 6 months is either a pathetic excuse in an attempt to shift blame, or scary they are that incompetent to figure it out in 6 months.
Either way what happened with the R520 and 7800 isnt what I would consider totally lucky. Nvidia planned well with the 6800 and 7800 series and have been hitting every goal since the NV30 debacle. At some point it goes from luck to great planning.
This latest incarnation was probably the most genius yet. They didnt win the speed title, and in some areas lost. But they have a cooler, less power consuming, and much cheaper to manufacture chip that competes. We are talking about a chip with half the die size competing with the competition. That isnt luck, that is great planning. Realizing they have an opportunity to simply hang while reducing production costs and increasing yields, they took it. Why pour money into the chip when you can do the above and focus on the next generation? While ATI fumbles around with a hot, power consumption hungry 48 pixel processor that apparently sits idle most of the time. Nvidia continues on its way and making nice returns for their owners.
While enthusiasts may not be "that" impressed, from a business perspective if you can compete with a product that costs half of your competition it gives you a lot of flexbility in what you want to do.
The OEM market is probably very interested in a lower power, cooler running chip. I am curious when Nvidia will bring out a 7900 variant for their laptops.
My take on the whole war is the R300 was a great hit, slapped Nvidia in the nutz, and got Nvidia back on track. ATi is one of those companies that I dont consider very well run. Before the R300 they werent doing anything but producing low perfoming, bug ridden hardware. After the R300 it is like they cant grasp their own market. When they introduce products in the channel they seem to focus on that speed crown and slowly integrate down the price points. I mean, have they finally killed off the 7500 they have relabeled to 9000,9100,9200? And how long did it take for them to finally get a top down SM3 product line?
The last 24 months havent been very impressive as the X800s either took months to show up or never did and the X1800 showed up late, hot, not overly fast compared to the competition, and was a short lived product cycle ala NV30. The X1900XT looks like a promising chip but I question if it isnt a generation ahead of itself with all that pixel processing power yet barely beats a competing Nvidia part that is almost half its size. And lets not delve into the idiotic design decisions of the X800 crossfire boards, which btw can you even purchase?
Mariner said:
Yep, but for the average guy a couple weeks wait is worth 100 bucks. If I had waited 3 weeks to buy my X1900XT I would've saved about that.
Anyway, I don't think that the poster I quoted was referring to a short term purchasing decision. If you read his quote again you see that he's drawing conclusions about yields, leverage etc, based on today's prices.
Mintmaster
Veteran
jb said:
Luck did not have very much to do with it at all. Sure, R520's late arrival hurt a bit, but ATI's lead in the R300 days were lost due to several serious mistakes in their design decisions._xxx_ said:
The biggest mistake of last gen was not including FP blending. It wasn't hard to see that HDR would be the next big thing, even as early as the R300 days. Surely they knew this was the most requested feature from devs. As we saw from the FarCry SM2.0b patch and SC:CT SM2.0 patch, developers were not going to shaft ATI simply for a lack of PS3.0 if the effect could be done in PS2.0 (which includes just about all effects in current games). The PS3.0 advantage could have been written off like NVidia's PS2.0a advantage in the FX days, but without FP blending, they really did have a practical feature disadvantage.
For this gen, ATI just went overboard with "PS3.0 done right". Even though I agree with the slogan, the need just isn't there right now. It really doesn't take a genius to extrapolate NV40's performance to a 300M transistor 90nm G71. I know the count is even less, but given that they used 320M on R520, ATI should have expected even higher performance from G71. For RV530, I don't know what they were thinking (even though I was impressed it passed the 6600GT with only 4 texture units). Still, look at a 110nm NV43 that's 150mm2, and you know that on 90nm NVidia will be able to boost speed at least 60% in the midrange.
Maybe it's all part of their long term plan, but ATI's economic disadvantage right now is no fluke.
Mariner said:
nV never used Low-k until the g71
. IMO it comes down to poor choice (planning) and execution by ATi. But now when nV screwed up with the nv30 the hurt didn't show up for at least a year, year and half down the road. ATi's screw ups started 6 months down the road and really hasn't stopped yet.Ati so far has had a lead in a general sense of speed up till the g70 starting with the r300. Now in the interm ATi wasn't able to secure more then 20% (giving them around a total of 50%) of nV's Desktop Discrete sales. This coupled with the the r300 technology and weakness on nV's part with the fx line they should have been able to take the entire market, there is no other competition in this segement.
Interesting we are talking about the scewed up libarary. Is nV using the same libarary? ATi didn't use this library for thier r500 so why did they use it for the r520 and kin? It must be something fairly important that they have incorporated.
trinibwoy said:
Making your own luck is a bit different, its more planned for, example from the art of war book, this is a true story, I can't remember the specific generals names though been a while back.
One general with 5000 soliders inside a city another outside with 25,000 soliders. The general inside told his men to leave the city at night. He kept a handful of soliders inside the city and dressed them up in rags and told them to sweep the grounds of the city. This general went to the top one of the towers and started playing his flute in the middle of the night.
The other general that was outside the city, saw this and wondered I know this man, he is either bluffing, but he couldn't be, in the end the general with 25,000 soliders retreated.
Was it really luck by the general inside the city that won the battle? In one aspect it is if the other general came in they would have crushed em. The other side of it is the general inside knew his oppenent better then the oppenent had cofindence in his own ability.
Well the moral of that story is that the general outside was pretty dumb
My point was that in a situation like ATi faced, the source of the problem is not the only factor (let's even assume that ATi had no way of anticipating such a problem). Once the symptoms surface, whose responsiblity is it to analyze the situation and identify the culprit? All I'm saying is how do we know that ATi needed all that time to identify the problem, maybe they could've done it faster, in which case they are also to blame.
My point was that in a situation like ATi faced, the source of the problem is not the only factor (let's even assume that ATi had no way of anticipating such a problem). Once the symptoms surface, whose responsiblity is it to analyze the situation and identify the culprit? All I'm saying is how do we know that ATi needed all that time to identify the problem, maybe they could've done it faster, in which case they are also to blame.
trinibwoy said:Well the moral of that story is that the general outside was pretty dumb
My point was that in a situation like ATi faced, the source of the problem is not the only factor (let's even assume that ATi had no way of anticipating such a problem). Once the symptoms surface, whose responsiblity is it to analyze the situation and identify the culprit? All I'm saying is how do we know that ATi needed all that time to identify the problem, maybe they could've done it faster, in which case they are also to blame.
LOL
True
Razor1 said:nV never used Low-k until the g71
Interesting we are talking about the scewed up libarary. Is nV using the same libarary? ATi didn't use this library for thier r500 so why did they use it for the r520 and kin? It must be something fairly important that they have incorporated.
Actually, as I remember it (and Beyond3D too), NV30 was originally planned for low-k but this process just wasn't ready. Hence the move to the standard 130nm process and the ensuing delays (the poor performance was due to a weak design in comparison to R300, however). ATI experienced a similar sort of production delay with R520/RV530 albeit caused by differing problems.
For me, ATI's problems started with R4x0. Although these were good chips with high performance, ATI lost mindshare due to the equally-performing NVidia NV4x chips which also supported SM3.0. The further delays to ATI's SM3.0 chips caused by the ensuing 90nm problems this meant that NVidia was effectively on their 2nd generation of SM3.0 parts long before a competitor arrived. NVidia is now reaping the benefits of their early and unchallenged SM3.0 adoption (as well as their more simplified implementation). G70 improved efficiency of their SM3.0 architecture and G71 (and G73) seem to have improved this further with higher performance from fewer transistors. This has put NVidia in a great position financially and of course it's the bottom line that counts!
I wonder, is it possible that NVidia also encountered similar problems with their 90nm chips to ATI? I suppose it is feasible that they used similar libraries to ATI although this wouldn't necessarily guarantee they would encounter the same bug. If there were technical reasons for the delay in NVidia's 90nm chips, they didn't really suffer as ATI didn't have products which could put real pressure on the NV4x series.
As I see it, ATI have suffered somewhat with feature-creep after R300. R400 slipped to become R500 and then R600 and it was necessary for ATI to produce R420 and it's derivatives which didn't really offer anything new technologically and NVidia took the 'features' lead.
However, we are approaching a inflection point of sorts later this year with the release of Vista. It seems entirely feasible to me that ATI's work on Xenos and USM will offer them a technological advantage in the next generation of chips, presumably R600 vs G80. In theory at least Xenos should have given them plenty of practical experience with their USM architecture so R600 could be considered their 2nd generation use of the technology with associated improvements in efficiency (I'd assume).
Whether or not G80 will compete with R600 or surpass it, I have no idea. But I am looking forward to finding out later in the year.
I don't consider NV30 to be an entire failure, if you consider it from a bit different aspect
It still was able to undercut the R300's price wise (NV35 and NV38 to R350 and R360, they were able to undercut the prices) and it offered a better midrange product, based on financials (FX5200 first DX9 low level card, FX5600 and 5600XT ended up in a nice price position after the FX5700 came, FX5700 put it to the 9600 series as well, I think ATI made a mistake in dropping the 9500). So, in reality nVidia seemed to do the same marketing and financials with NV3x that it pulled off with NV4x and G7x...
nVidia sometimes does have the performance crown for some time (the 6800 Ultra, the 7800GTX, both at their launches were the fastest thing on the block) but they always manage to keep prices down in time to compete with ATI, and they gorge out a nice midrange market (FX 5200 still sells today, 9200 doesn't really, and not at a good price poitn, 6200 and 6600 still ship today (so do the FX 5500 and 6500 series) yeah it's a ton of models to ship, but it keeps the sub $100 range flooded with NV parts)
But if you look at ATI's high end parts post R300
9800XT
X850XT
X1800XT
and X1900
they surpass nV's offerings (5900, 6800, 7800, 7900, etc) by a bit, but they manage to run so much cheaper...while a performance tuner can consider a 9800Pro a bargin in it's day compared to an FX 5900 Ultra, the average Joe at compusa won't because the average Joe won't OC their 9800 and take advantage of the chip's potential, they'll consider their 5900 Ultra a good buy for costing $50 to $100 less and offering nearly the same performance, so what that it's bigger?
I don't know much of NV's history regarding NV-ATI competition around the time of the Radeon 7000 and 8000, however I do know some of the scraps between 3dfx and NV...NV managed to do the same thing to 3dfx it's doing to ATI, it would ship features seen in the future, first, and get them done mostly right...
Correct me if I get models wrong here, but with Voodoo3 3dfx didn't have the color depth of, I wanna say TNT2 but it might be the GeForce 256, while the Voodoo3 was still fast
Which is comparable to say, the 6800GT and the X800Pro...both are fast cards, however the 6800 can hold it's SM3.0 and FP32 over the X800 and match it in prices...
It repeats, nVidia cycles products through in a similar manner, keeping prices under their competition while putting out more features where possible, either as an advertising plug to the common guy, or to make an enthusiast consider a slightly slower product in exchange for more features (to make it seem like more of a value: 90% of the power, 80% of the price, 120% of the feature support, that kind of deal)
I'm not sure you can draw direct correlations, because ATI hasn't made some of the same mistakes as 3dfx, they aren't ignoring the consumer market and pushing into the OEM market soley, which is what 3dfx was reported to be attempting to do in 1999 and early 2000...that's just my 2 cents...
It still was able to undercut the R300's price wise (NV35 and NV38 to R350 and R360, they were able to undercut the prices) and it offered a better midrange product, based on financials (FX5200 first DX9 low level card, FX5600 and 5600XT ended up in a nice price position after the FX5700 came, FX5700 put it to the 9600 series as well, I think ATI made a mistake in dropping the 9500). So, in reality nVidia seemed to do the same marketing and financials with NV3x that it pulled off with NV4x and G7x...
nVidia sometimes does have the performance crown for some time (the 6800 Ultra, the 7800GTX, both at their launches were the fastest thing on the block) but they always manage to keep prices down in time to compete with ATI, and they gorge out a nice midrange market (FX 5200 still sells today, 9200 doesn't really, and not at a good price poitn, 6200 and 6600 still ship today (so do the FX 5500 and 6500 series) yeah it's a ton of models to ship, but it keeps the sub $100 range flooded with NV parts)
But if you look at ATI's high end parts post R300
9800XT
X850XT
X1800XT
and X1900
they surpass nV's offerings (5900, 6800, 7800, 7900, etc) by a bit, but they manage to run so much cheaper...while a performance tuner can consider a 9800Pro a bargin in it's day compared to an FX 5900 Ultra, the average Joe at compusa won't because the average Joe won't OC their 9800 and take advantage of the chip's potential, they'll consider their 5900 Ultra a good buy for costing $50 to $100 less and offering nearly the same performance, so what that it's bigger?
I don't know much of NV's history regarding NV-ATI competition around the time of the Radeon 7000 and 8000, however I do know some of the scraps between 3dfx and NV...NV managed to do the same thing to 3dfx it's doing to ATI, it would ship features seen in the future, first, and get them done mostly right...
Correct me if I get models wrong here, but with Voodoo3 3dfx didn't have the color depth of, I wanna say TNT2 but it might be the GeForce 256, while the Voodoo3 was still fast
Which is comparable to say, the 6800GT and the X800Pro...both are fast cards, however the 6800 can hold it's SM3.0 and FP32 over the X800 and match it in prices...
It repeats, nVidia cycles products through in a similar manner, keeping prices under their competition while putting out more features where possible, either as an advertising plug to the common guy, or to make an enthusiast consider a slightly slower product in exchange for more features (to make it seem like more of a value: 90% of the power, 80% of the price, 120% of the feature support, that kind of deal)
I'm not sure you can draw direct correlations, because ATI hasn't made some of the same mistakes as 3dfx, they aren't ignoring the consumer market and pushing into the OEM market soley, which is what 3dfx was reported to be attempting to do in 1999 and early 2000...that's just my 2 cents...
Razor1 said:
No, luck happens in business and it being "rare" is a matter of opinion. Yes there are lot of opertunities that are nothing more than hard work and great planning. And then every know and then you catch a break (ie lucky).
There is no way around it that NV cought a major break with R520 being so delayed, and thats how they got lucky. Had the R520 been on time many things could be vastly different today. I am not saying that all of this sucsess is luck. No its very great work by NV on many fronts. But yes they did get catch a break with R520 delay.
Mintmaster said:
See above. Having the R520 being as late hurt a lot more than a bit....
jb said:
How can you call it luck, when nV probably already had something else scheduled then a nv47 to go up against the r520? as stated before and know for a while now the nv47 was the refresh that should have gone up against the r480. Its not luck, its planned and executed better.
nV has been doing this for quite sometime now. Another strategm cutting off trade routes. Supply of money. Money comes from midrange sales, nV has had the ground from the gf 6 and up. No matter how you look at it, luck is made, it is not there. Anther area is public preception, ATi can't penetrate this at all. If its purely a whismisical chance, its better to play the lottery, at least you loose a few bucks, and not a few million. The moment someone in business says it luck, thats the end of it. They stunt thier vision.
Last edited by a moderator:
Similar threads
