Next NV High-end
- Thread starter Geo
- Start date
It doesn't say that. The G72 remains a low-end chip, afaik, although either side of the story could be misinformation. As for the "multiple chips on one board", I would personally suspect the workstation market to be their aim with that, but I could be wrong.geo said:
Uttar
trinibwoy said:
Sure, but if that doesn't work they can always counter through prices. Or move their chips one step lower on the comparison chart
(like, they'd put GTX as a competitor for X1600 instead of GT, the new ultra/whatever-the-name-thingy against 1800XT etc.).Anyway, ATI should get the R580 out ASAP or they'll look REALLY bad in a few months.
http://www.theinquirer.net/?article=26929
Huh?? What's this say about the timing for NV's 90nm mid-range?
Huh?? What's this say about the timing for NV's 90nm mid-range?
It is just a NV41 on 110nm process (with 1 VS less). Looks like a "gap-filler" card to combat GTO/GTO2's success.
serenity said:
But why would you need a gap filler unless your new mid-range either isn't competitive (unlikely) performance wise or won't be showing up for at least another two months? Are we expecting it to be that late? Mid-December or later? I guess that's not too bad if X1600 really hits Nov. 30th. Unless the "or later" is substantial.
Er, only if it doesn't compete with their current hardware. That is to say, for nVidia to do this, they'd have to have the expectation that the 90nm part would be released in a completely different price bracket from the current G70. Only then would it make sense to plan to release the two parts in such quick succession.BRiT said:
But since nVidia already released the G70 at an unprecedented price level, it seems extremely hard to believe that a higher-performing refresh would be priced still higher at release. So any 90nm version of the G70 that is meant to be a high-end part simply wouldn't be in the plans until at least next spring.
I hope you folks aren't assuming just because the supposed "G72" might be manufactured at low-k 90nm that it "has" to be a high end part...
That's a pretty shortsighted theory IMO. Any IHV would try to milk first out every inch of possible revenue from a product line before they move on to a new one (exception would be complete failures). You're sure you're not thinking of something that results rather in loss than profit?
Mainstream and budget first to slowly phase out lower end products IMO and then a new high end part. Prior to that we'll see the R520-availability bickering threads first anyway hm?
That's a pretty shortsighted theory IMO. Any IHV would try to milk first out every inch of possible revenue from a product line before they move on to a new one (exception would be complete failures). You're sure you're not thinking of something that results rather in loss than profit?
Mainstream and budget first to slowly phase out lower end products IMO and then a new high end part. Prior to that we'll see the R520-availability bickering threads first anyway hm?
Agreed.geo said:
Whereas it seems ATI's X800 GT/GTO/GTO2 SKU's were created mainly to offload excess X8x0 inventory before R5 SM3.0 parts hit and also to put pressure on 6600GT sales, the creation of the 6800 GS SKU (if indeed it becomes a volume SKU) would lend credence to the suggestion that nV's mainstream (7600?) and perhaps low-end G7x 90nm parts are still a few months off. I wouldn't think nV would have an inventory problem with NV4x given their strong position this year, so creating the 6800 GS to offload excess NV42's wouldn't seem likely unless nV have made some misjudements with their order forecasting.
I suppose this delay from the originally surmised launch window (Q4 05) for mainstream G7x doesn't necessarily indicate production/yield issues with nV's 90nm ramp, it could be that nV don't feel threatened by the X1600 and X1300 performance (and limited availability, at least with X1600) for the time being and perhaps want to build up inventory for a wide launch in Q1 06, or perhaps as Dave or someone else indicated, nVIDIA are being constrained by fab space, with ATI and Microsoft using most of the 90nm lines at TSMC -- I believe someone mentioned that nVIDIA are working with or considering utilising Charter and IBM for 90nm, as well as TSMC. Available production capability must be a factor this cycle, with ATI also using UMC for R5xx. Alternatively, maybe nVIDIA are having problems with 90nm and low-k?
Bear in mind that NVIDIA are probably trying to run through inventory of NV41/42 much like they are NV40, so its not necessarily a given that releasing this may be an indication that the 90nm products aren't coming - it could equally be an idication that they are coming and they want to get rid of these parts beforehand; much like ATI running through R430/R480 inventories as much as possible before the release of R5xx. I'm not saying this is the case, but its certainly something to consider.
