Penstarsys Speculates on upcoming nVidia products...

I found this blurb on AnandTech, but unfortunately, I can't read the whole article at PenstarSys. Didn't they used to be some sort of affiliate of B3D? (Or shared server space, or something?)

http://www.penstarsys.com

have a feeling that the NV-28 chip from NVIDIA will not be an advanced refresh from the current NV-25 chip. In fact, it is looking to be a cut down version of the NV-25, which will give the mainstream and low end full DX 8.1 compatibility for NVIDIA. Unlike the ATI Radeon 9000, which has 4 pipelines with a single texture unit per pipe, it appears that the NV-28 will only have two pipelines, with 2 texturing units per pipe. Instead of the dual Vertex Shader units on the NV-25, the NV-28 will feature one. All of the other parts look to be about the same (LMA II, AA engine, nView), but I am pretty sure the Video Processing Engine will be added to the NV-28 (such as the one in the NV-17). This chip will probably be as fast overall as the Ti 4200, but it will be much, much cheaper to produce. The NV-28 would weigh in around 36 million transistors, while the NV-25 core is around 64 million transistors. Die size would be approximately half of what the NV-25 is, assuming that NVIDIA will use the .15u process for this new chip. If NVIDIA goes ahead and uses the new .13u process, then the die sizes will be nearly 1/4th that of the current NV-25. This would make each chip substantially cheaper to produce, and they would be able to easily clock to 350 MHz or so. The NV-28 would of course also be a AGP 8X part. The die shrink and redesign would also allow the NV-28 to be used in high end notebooks, providing competition with ATI's recently announced Mobility Radeon 9000. It would then include all of the power saving tricks that NVIDIA has used in the NV-17M chip. The NV-28 appears to be the mainstream warrior for NVIDIA this coming season, and will provide excellent competition to ATI's lineup.

Talk about something that goes completely against conventional speculation! NV28 at only 30 mil transistors? I thought it was pretty well documented by now that it's at least 60 mil, and likely 80?

Besides, I don't know how he reaches the conclusion that a 2 pipe, 2 TMU/pipe version of the GeForce4 Ti would be "as fast overall" as the Ti 4200, which is a 4 pipe 2 TMU/pipe....

Nothing really makes sense to me...
 
The guys makes sense, but his "speculative" thoughts about NV28 don't...

Come on guys, even on that messy slide it was obvious that it's not 30+ million transistors!

And 2x2? That surely won't be on par with gf4ti4200...
 
McElvis said:
Not unless it's on 0.13 and clocked at very high speeds.

The CEO said several times that the only 0.13 part will be NV30 and the latest rumours suggest that NV28 will only be an AGP 8x part and the clockspeeds will remain the same...

I sure hope NV28 will be more than just "AGP 8x part"...
 
I'm personally open to any and all rumors from reputable places. If we get enough of them, they'll eventually line up.

Of course, we're obviously no where near 'enough' at this point. (clarify: to get them to line up)
 
alexsok said:
McElvis said:
Not unless it's on 0.13 and clocked at very high speeds.

The CEO said several times that the only 0.13 part will be NV30 and the latest rumours suggest that NV28 will only be an AGP 8x part and the clockspeeds will remain the same...

I sure hope NV28 will be more than just "AGP 8x part"...

Good point.

His 'guesses' on a mobile NV28 seem to agree with Tom's Hardware though. And a 2x2 on 0.13 process would be a better option for a mobile (in regards to power usage) than a full blown GF4
 
Yes, I think the mobile rumor nearly eliminates the normal GF4 design on 0,15um(exept it is only for portable desktops ;) ) so this rumor does not sound silly, and at the moment nvidia lakes a DX8 lowcost/notebook chip (like the RV250).

If nvidia manages to make this design to run with 400-500MHz maybe around an enhanced 0,15u process, I would think that it could reach the speed of an GF4Ti. Only the transistor count seems to low for me (atleast if they dont reduce the caches drastically. If they do it, I wonder how they reach the speed)

@alexsok as far as I remeber the 3 looked like an 8, so I would not eliminate that theory.
 
@alexsok as far as I remeber the 3 looked like an 8, so I would not eliminate that theory.

I assume you're referring to that messy pic of nvidia's roadmap?

Yeah, it was kinda confusing there, but as one guy here pointed out not long ago, there was a presentation on T.V and the slide was clearly visible there and the numbers were: 81 million for NV18 & 86 million for NV28 and the CEO repeated these numbers.
 
OK, if the high number is really true:
Maybe nvidia tries to sell the both of my assumed low-cost chips (nv18 and nv28) alone and combined with a cpu "chipset"(like the nforce) on die. That would also fit in the high tape-out count.
Marketing people would choose the higher numbers (you know, the higher the better) and show both transister counts as "nforce" versions. That would at least make much more sense for the nv18, and maybe it is also true for the nv28.

EDIT:
Additional hint: if the nv28 would be a "high end" chip, I think nvidia would try to hype it much more. And it is very quiet about the nv28. And only adding agp8x does not make much sense for me.
 
http://www.reactorcritical.com/#l1224

Nvidia showcased GeForce4 MX440 with AGP 8x support at “AGP 8x Plugfestâ€￾ held by VIA Technologies this Wednesday in Taipei. Later the company will unveil their GeForce4 Ti4200 and Ti4600 as well as Quadro4 line of professional products with AGP 8x support. All the graphics solutions should come sometimes in October timeframe and there will be no clock-speed differences compared to the current graphics cards.

This Fall Nvidia is planning to discontinue their GeForce4 MX460 and GeForce4 Ti4400 since numerous companies already launched cheap and advanced versions of the GeForce4 Ti4200, affecting the sales of the mentioned video solutions. Instead of the GeForce4 MX460, the Special Edition of the GeForce4 MX440 is going to be launched. It is quite easy to guess that it will resemble its predecessor with “460â€￾ index in terms of faster memory use. As for the successor of the GeForce4 Ti4400, graphics cards makers now start to announce advanced models based on the GeForce4 Ti4200 and utilising BGA DDR SDRAM.

Granted, much of that should be treated as rumour, but I have my suspisions about NV28 being a cut down GF4 Ti for the low end becuase that would effectively make GF4MX obsolete - in which case whats the point to NV18 and showcasing an MX with AGP8x support?
 
Granted, much of that should be treated as rumour, but I have my suspisions about NV28 being a cut down GF4 Ti for the low end becuase that would effectively make GF4MX obsolete - in which case whats the point to NV18 and showcasing an MX with AGP8x support?

Now that I think of it, it actually makes sense!

So they will kill the gf4mx (dx7 out of here - finally), and make a cutdown version of gf4ti marketed as nv28 on 0.15 (donno pipelines amounts nor clock speeds, don't even have an assumption).

This way, nv28 will effectively be capable of facing R9000pro and it's self-price will be much much lower than gf4ti's and it will be targeted for the mainstream market or even lower than that (who knows?).

If that indeed happens, the mainstream market will finally move from the dx7 era to dx8 era and there will be competitors in forms of: NV28, RV250, XP4 and maybe something else...
 
I don't think you quite got what I was saying - if they were to kill off the MX then why bother refreshing it...?
 
DaveBaumann said:
I don't think you quite got what I was saying - if they were to kill off the MX then why bother refreshing it...?

Oh, yeah, I missed that line...

Donno why, sounds useless to me to refresh a dx7 part when the rest of the industry moves the mainstream and lower markets to dx8 era...

What I wrote above sounds the most reasonable (at least to me it does)...
 
Why would they be showing an nForce product (a competitor to Via's main market space) at Via's own 'plugfest' show?

[edit] More info on the Via event:

http://www.via.com.tw/en/Digital Library/pr_agpplugfest.jsp

Taipei, Taiwan, 27 August 2002 - VIA Technologies, Inc, a leading innovator and developer of silicon chip technologies and PC platform solutions, announced that it will host a Plugfest event to validate the AGP 8X performance of VIA Apollo P4X400 and KT400 chipset based motherboards, that will take place on Wednesday 28th August, 2002 at the Taipei International Convention Center, Taiwan.

Opening the event on Wednesday, VIA will be joined by Microsoft and leading graphics vendors ATI and NVIDIA in highlighting the importance of AGP 8X to the industry. This will be followed by extensive validation testing of premium performance Socket 478 and Socket A motherboards available from a broad spectrum of manufacturers, including Abit, Albatron, Acorp, Aopen, Azza, Biostar, Chaintech, DFI, ECS, EpoX, Giga-Byte, Iwill, Jetway, Micro-Star, Shuttle, Soltek, Soyo and Tyan.

Delivering 2.1GB/s of bandwidth to the graphics engine, the new AGP 8X standard enables improved image quality and 3D performance in the highest resolution conditions, and looks set to be widely adopted by the industry. Underlining VIA's commitment to open industry standards and technical compatibility, the AGP 8X Plugfest will bring together key players to validate the latest hardware supporting this new feature.

"VIA remains committed to driving technical innovation, and working with our industry partners to facilitate - not impose - wide and rapid market adoption of beneficial new applications such as AGP8X," commented Richard Brown, Director of Marketing, VIA Technologies, Inc. "AGP 8X support enables end-users to realize unequalled image quality and performance, so this event is vital to ensure greater product compatibility, and we are delighted that key players are joining us to achieve this goal."

"VIA's Apollo P4X400 and KT400 performance chipsets both support the next generation graphics capability of AGP 8X, and OEMs and system integrators are keen to incorporate this powerful combination into their top of the range systems," said Chewei Lin, Vice President of Product Marketing at VIA Technologies, Inc. "The AGP 8X Plugfest is aimed at the harmonization of standards within the industry, so facilitating a more rapid uptake of this high performance feature."

Seems to me that NV wouldn't be showcasing nForce products there, but graphics boards that can 'plug' in to the Via 8X boards.
 
There are way too many speculations going...

So we know the trainsistor counts (NV18:81m, NV28:86m) but we don't know what else will be added to the NV17 & NV25 cores besides AGP 8x, or maybe a cutdown will occur and not an addition?

If the NV18 wouldn't be announced and shown in Nvidia's roadmap, it would be safe to assume that NV28 is a cutdown NV25 core for the mainstream or lower markets... but, what about the rumours NV28 should replace NV25? God, that's getting hard...
 
So we know the trainsistor counts (NV18:81m, NV28:86m) but we don't know what else will be added to the NV17 & NV25 cores besides AGP 8x, or maybe a cutdown will occur and not an addition?

No, you don't know that. You know that the CEO has said that, but that doesn't necessarily mean it to be the case.
 
And its not beyond the realms of possibility that whoever wrote those slides and his notes could have got it wrong?
 
Another report from Digitimes:

http://www.digitimes.com/NewsShow/Article.asp?datePublish=2002/08/30&pages=02&seq=7

As for the major graphics chip supplier Nvidia, industry sources said that the chip designer is expected to introduce its AGP 8x-supporting NV18 and NV28 chips in October, and roll out its new-generation NV30 in time for the Christmas season. The basic architecture of the NV18 and NV28 is not very different from that of the company’s previous NV17 and NV25, but the launch may be able to help Nvidia keep its market share from declining further amid strong competition from ATI and Intel.
 
Back
Top