POLL: nVidia's Silence on GT300

Page 8 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Keysplayr

Elite Member
Jan 16, 2003
21,211
50
91
Originally posted by: Qbah
Originally posted by: Keysplayr
Originally posted by: Qbah
The drawing you show here as proof was done by someone who incorrectly read the slides showed earlier. While it may be possible the physical layout will be like AMD's slide (and not like on the private sketch that's different from the official slide anyway), it's all about the logical pathways on the diagram and in the physical chip. Since we still don't know how the chip looks like inside in detail and the logical distribution on the slide is fully parallel, those claims are false.

As much as AnandThenMan's statement was a bit blunt and straight to the point, I share his view that right now it's grasping at straws when trying to discredit ATi's new design.

And I would like to apologize to Keys for my harsh words earlier. They were unnecessary and do not represent my way of discussing things. Had a few bad weeks recently and I get annoyed easily. Again, sorry.

We all have our off days/weeks. Thanks. And it would be great if everyone just understood that I am NOT trying to discredit ATi's (AMD's) new design. I didn't say there was anything wrong with it, not whatsoever. I was just making an observation of how it appeared from the schematic pic. Every single feature of R7xx was doubled. Exactly doubled. 800 to 1600sp. 40 to 80 TMUs. 16 to 32 ROPs. The only difference being a 256-bit bus across the board. Telling me that if they DID slap two cores side by side, they are not the true R770 cores, but more like two R740 cores (128-bit memory interface) but with the full 800sp 40TMU 16ROPs of the R770. Like a hybrid. Something that if made today would be called something like a HD4790.
Saying the logical pathways doesn't really tell me that there aren't two separated GPU's in the R8xx core.

Looking at the second photo Kakkoii linked to, tells me that there are two Command Processors, Two Setup Engines, Two interpolators and Two Ultr-Threaded Dispatch Processors. Is this schematic inaccurate? Are there only one of each? Because is does not look like these items are "connected" via logical pathways as you have mentioned earlier. It looks like the only time data actually gets "merged" from the seemingly two cores, is when the data reaches the high Bandwidth Crossbar/Shader Export.

If this was TRULY a completely monolithic core, I would expect to see only one Command Processor, one Setup Engine, One interpolator, and one Ultra-Threaded Dispatch Processor across ALL SIMD Blocks and TMU's.

Again for emphasis, I am not discrediting this design, at all. If performance is there, and it seems like it is, nobody really cares if it's a single monolithic core or two setup side by side. This is just for some interesting discussion while we are waiting for the big "GT300 vs. R8xx" wars coming.

And even if I was discrediting the design, nobody should really care one way or another. When it comes down to the wire, nobody cares. They just care how it performs. Hell, look at Core 2 Quad. Do people care that it is essentially 2 Core 2 Duos in one package? I don't think so.

:thumbsup:

Original slide is here. One command processor, one ultra-threaded dispatch processor that goes to all SIMD blocks. No mentioning of the Setup Engine and Interpolator on the official slide.

Anyway, I can see why one could think like that. And it can go both ways in the end I guess. But based on what's on the slides, it's one monolithic chip. Either way:

Originally posted by: Keysplayr
When it comes down to the wire, nobody cares. They just care how it performs.

But then by which pic can we actually go by? They seem to conflict. I think a nice clear actual die shot might be the tell all.
 

MODEL3

Senior member
Jul 22, 2009
528
0
0
I think it is a monolithic design.
It doesn't make sense other wise (without complete memory sharing the scaling will be bad and would require driver optimizations per game...) (and the design seems similar with the RV770 so i don't think complete memory sharing is a very probable scenario)
For me it's just a marketing slide...
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Originally posted by: Keysplayr
But then by which pic can we actually go by? They seem to conflict. I think a nice clear actual die shot might be the tell all.

Originally posted by: MODEL3
I think it is a monolithic design.
It doesn't make sense other wise (without complete memory sharing the scaling will be bad and would require driver optimizations per game...) (and the design seems similar with the RV770 so i don't think complete memory sharing is a very probable scenario)
For me it's just a marketing slide...

Another thing we can apply to assist in bounding our creativity on the possible explanations/outcomes here is that we know AMD likes to harvest their chips as a means of cost-reduction and product differentiation.

So the question to ask ourselves is does one arrangement lend itself more readily to die harvesting than the other arrangement?

And if it does, then what rational argument could we envision would be used to still go with the other (less amenable to die harvesting) implementation?

Hypothesis, null-hypothesis, etc.

FWIW I can think of one nice intrinsic advantage of going with a dual-core design - power consumption. If a game doesn't need both cores fully ramped up to push those frames then it could idle the underutilized resources to reduce heat and noise. Could be useful in those situations where anything over 60fps is wasted computing on the GPU's behalf.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,792
1,512
136
Originally posted by: theAnimal
Originally posted by: lopri
Why does it need three 6-pin connectors?

Why indeed...an 8-pin connector is the same as 2 6-pin.

No, an eight pin is the same as a 6-pin with two grounds. A six pin connector can pull as much power as the eight pin can if pushed to it, it's just not rated to do it safely. 2 six pin connectors have the same 'rating' as one 8 pin connector, but they can push more watts in the real world.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |