Nvidia GT300 core: Speculation
- Thread starter Shtal
- Start date
- Status
aaronspink
Veteran
aaronspink
Veteran
This is incorrect. Server memory does not store CRCs as it won't allow you to correct. The store various ECCs which generally are variants of hamming code.
OpenGL guy
Veteran
Yes, I was attempting to use the same terminology as the original poster.
aaronspink
Veteran
On a related note, does anyone else get the impression Theo has no idea what he's talking about sometimes?
real ECC/SECDED for graphics without giving up something is going to be hard. I'm not convinced there is enough of a market for x36/x18 GDDR and the granularity really makes doing anything else in the actual memory system difficult.
I've yet to see an example of the opposite...
aaronspink
Veteran
That works.
I find it interesting that people haven't started speculating why they decided to add the CRC to GDDR5.
Am I the only one that has that impression with many other online "journalists" (with very rare exceptions)?
In general it isn't, but the one they used is.
Hmmmm so "half the planet" believes it has severe hw problems and it won't appear prior to Q2 10 and the other half believes it'll demo as early as September. NV has managed to close all possible leaks since the beginning of this summer, so I'd really love to know where either/or get their supposed infos from. Maybe I should start chewing eucalyptus leaves, hug a tree and claim to be an australian "source". I bet it'll work
I didn't know Aussie sources were koalas...Hmmmm so "half the planet" believes it has severe hw problems and it won't appear prior to Q2 10 and the other half believes it'll demo as early as September. NV has managed to close all possible leaks since the beginning of this summer, so I'd really love to know where either/or get their supposed infos from. Maybe I should start chewing eucalyptus leaves, hug a tree and claim to be an australian "source". I bet it'll work
Hmmmm so "half the planet" believes it has severe hw problems and it won't appear prior to Q2 10 and the other half believes it'll demo as early as September. NV has managed to close all possible leaks since the beginning of this summer, so I'd really love to know where either/or get their supposed infos from. Maybe I should start chewing eucalyptus leaves, hug a tree and claim to be an australian "source". I bet it'll work
If you take Charlie's fast-track story you'll end up with a mid-September availability of those chips. in the best of all possible cases they could have GT300 up and running at the time the HD5 series goes on sale. Retail availability still won't be there in 2009, but a demo in their own September event has to be achieved if they want to have the slightest change at debunking NV30=GT300.
On the other hand, there's still this group that has GT300 as a radical new design that's much smaller than GT200 and that has been secretly implanted in a lot of devices to take over the world, once again.
I don't want to take any Charlie's or any Theo's stories for any kind of reference. Just read back a couple of posts and you'll know why.
On the other hand of the other hand it's still uncertain if D12U is a rather "revolutionary" or "evolutionary" design. If it's the latter history will just repeat itself; if it's the first there's more than one individual and his typical entourage that will have some explaining to do. It's in the best interest of the consumer just in case you missed it that we see as much progress as possible and I personally don't care where it comes from since I don't have either any interests, agendas or ulterior motives. Don't ask yourself; ask those who have.
Ok. Your thoughts? :smile:
An alternative to ECC for Tesla?
I was under the impression that most of these so called journalist were like this. Heh
It's just there so they can go closer to the edge as far as signal integrity is concerned.
- Status