News: The 7800gtx isn't really the G70....

[DilTech]

Guys, this is 100% my story, no hardware site has it posted, so if there's any complaints, take it up with me directly.. I'm still digging for more info, so this will be updated... Either way, it IS news, so don't move it!

First, I'll start you guys off with this...
http://www.xbitlabs.com/news/video/d...110034348.html
http://translate.google.com/translat...ial_s%26sa%3DN
http://translate.google.com/translat...ial_s%26sa%3DN
As you can see from that, this summer, ORIGINALLY, the NV47 was suppose to release, being 24 pipelines, 8 vertex shaders, and 110nm design. The chip was a MASSIVELY revamped NV40.... The chip was then supposidly cancelled so that NVidia could concentrate on the G70, a core that was suppose to be in the works along side the NV47, however, it turns out the NV47 was never actually cancelled... The 7800gtx IS INFACT THE NV47.

Ladies and gentlemen of the jury, exhibit's A and B are the attachments you'll see at the bottom of your screen. Notice that both Rivatuner AND NVidia's own Development software BOTH call the core the NV47.. If this core was infact the G70, wouldn't NVidia have had it hardcoded to say it was rather than the NV47? Didn't NVidia say this so-called NV47 was cancelled?

Exhibit C...Please look at the picture of the G70 core very closely while reading this one...
http://www.digit-life.com/articles2/...g70-part2.html

According to its designation, the chip is manufactured or packaged in Korea. That's confusing. It's well known that TSMC has its own facilities for packaging dies, that's why the die should have been stamped TAIWAN, if manufactured by TSMC. There is a packaging plant in Korea, but it mostly processes dies from IBM. Has the American blue giant taken up NV47\G70 manufacturing again? The matter is left open.

Now, We all know that TSMC produces the G70 core, right? Well, TSMC DOESN'T HAVE FABS IN KOREA! IBM uses korea, and IBM isn't making these cores... What does this have to do with anything you ask? THE STAMPS ARE ADDED BY NVIDIA AFTER THE CHIP IS PRODUCED! NVidia messed up by leaving it to say Korea instead of Taiwan. A mistake only very few people would notice, as most people don't take the heatsink off their videocard, and even less know where TSMC's fabs are.

Now, there's the news part out of the way.. The 7800gtx is NOT the G70, but the NV47, which officially leads us to the speculation part...

Why would NVidia lie about the core being the G70, and even more-so, what would they have to gain from such a claim? They could've easily told everyone it was going to be the NV47, and no one really woulda cared, as long as the performance numbers are what we're looking at now(and they are SWEET!), so why lie about it?

My theories on this are the following...
1.) The actual G70 wasn't finished in time, therefore NVidia went with what they already had done, the NV47, while they finish up the G70, making it either 90nm, 32 pipes, or both. Giving ATi a false sense of hope that the NV47/g70 is the best they have at the same time.
2.) They're sittin on the G70, as they knew the NV47 could make some major sales, realizing there was huge delays with the R520. Thus, changing the G70 name to G71, and using the NV47 to make ATi show their hand.
3.) There never was a G70, NVidia just threw out the name, replacing the NV47 name, to really throw a curveball at ATi... Increasing the pressure on them, essentially, bluffing.

Regardless, NVidia is going thru ALOT of trouble to hide the core's true name if it's number 3...So 1 or 2 is the more probable answer.

Let me know whatcha think guys, I'm going to continue to dig up info and see if I can get to the bottom of this. I've also emailed NVidia about this, and if they give me an official answer I will post it here asap... Again, the part of the 7800gtx not really being the G70, but the NV47 with a fake stamp on it IS news, so please do not move or delete this thread, it WILL be updated.

[.sentinel]

That is a lowball trick if it is right. This is why I buy last-gen cards becuase of scandals like this.

[DilTech]

Oh, it's right... NVidia's developer control panel AND rivatuner both state the 7800gtx is infact the NV47, and read any ATi core for, they all state taiwan for place made, because that's where TSMC's fabs are... Why does the G70 say korea when the only GPU people who make cards there are IBM, while NVidia specifically stated they're going thru TSMC this round.

The proof is all laid out in the post.

[k0nsl]

Smart move by Nvidia if what you say is true. This is also why I just bought a 6800GT - maybe the "real" G70 will be unveiled when ATI launches their new product. I certainly hope so!

-k0nsl

[[XC] leviathan18]

mmmm so this why they could have in masses for launch date smart move they used the project they cancelled to gain some time for g71

so they have an As under the sleeve or maybe they were having the same problems as r520 with the pipes and 90nm process

[DilTech]

I'm glad to see people aren't taking this story as flamebait, or complaining calling it a theory. Really nice to see that members here actually read instead of skimming thru a couple of lines and calling it BS, when honestly, it's clear as day that it's all true.

The question remains though, why is NVidia going thru so much trouble as to make it out to be the G70 when it's really the NV47?.. I have emailed NVidia, as a News member of this forum, asking for an answer. Hopefully I get an actual reply and not an automated response, considering if I do get a response it WILL be posted on here asap. If I don't get a response I'll post this thread on their forums until I finally do get an answer.

I *WILL* get to the bottom of this... If anyone else finds ANYTHING out, please post it in this thread. Thank you in advance.

[[XC] leviathan18]

i dotn know perhaps they just tape out the nv47 as a g70 cuz they were having troubles like ati right now

so the real g70 will be the g71 but they are dead silent with the g71

[metro.cl]

weird thing to do

[freecableguy]

I believe it. I think there is something to this story!

[Mysterfix]

If it turns out to be all true it's a brilliant plan. They bate ATI with the NV47 while they secretly work on the G70 to combat the R520.

[Ugly n Grey]

Is it not possible that some components of the cancelled NV47 were carried over to the G70? I mean, they had some good engineering obviously, why start from scratch? I think the G70 IS totally based off the NV47 core that was never actually built. Why change what the chip is reported as in development tools? It's probably too much work for no real gain considering they were half way done when they stopped.

That makes logical sense, moreso in my opinion than than having unused technology lying around, the board of directors would fire them all for wasting bucks otherwise.
UnG

[[XC] leviathan18]

so the true new gen is the g71 not the g70

[freecableguy]

Also, maybe the missing key is that ATI already knows this! That's why they don't feel rushed to release Fudo. They KNOW that it will crush the NV47/"G70" and therefore don't need to take the bait. That would certainly explain the reports of the R580 core being tapped out. NV47, followed by R520, QUICKLY (before Xmas 2006) followed by the "G71" and R580!

-FCG

[[XC] leviathan18]

i want to see the mainstream product nvidia has to release to compete against ati

[cadaveca]

um...rivatuner identifies my 7800GTX as NV47, revision A1....i've seen mostly A2 chips on the net....mine is most likely a real nv47, as it is rom strapped from device ID "0090" to "0091"....0090 was the NV47, i think, and "0091" is the string to name is 7800GTX.

[DilTech]

Also, maybe the missing key is that ATI already knows this! That's why they don't feel rushed to release Fudo. They KNOW that it will crush the NV47/"G70" and therefore don't need to take the bait. That would certainly explain the reports of the R580 core being tapped out. NV47, followed by R520, QUICKLY (before Xmas 2006) followed by the "G71" and R580!

-FCG

Then the G80(which is NVidia's counter to the R580), which is also reported to be taped out, and was not even 2 weeks after the 7800gtx launch. Of course, ATi have already stated they will not release the R580 until next spring/summer, as they cannot afford to take sales away from the R520 due to the sheer amount of cash sank into that project. It has to be able to sell for top dollar for it's timeline, or they're going to take FAR more losses than they will gains on that card. Again, ati is already being sued for lying to it's shareholders, they cannot afford to do the same again.

Also, UnG, even if you were right(albeit the G70 specs match the NV47 specs TO A T!), that STILL wouldn't explain the obviously fake stamp on the G70 stating it's made in korea when TSMC's fabs are in taiwan. NVidia are really trying to cover this all up here. Also, it's not the code in the developer tools nor Rivatuner that's the issue, it's the hardcore on the core of the NV47/G70 itself that's reading as NV47. Using the same components wouldn't explain why they're being hardcoded as NV47 in the slightest. Also, as I stated, the specs match up to a T, from the process, to the vertex shaders, to the pipelines... Every review site has said it highly resembles the NV40, which the NV47 was a massively revamped NV40... This IS the NV47, not just based on it, but it actually IS that core.

[cadaveca]

um...rivatuner identifies my 7800GTX as NV47, revision A1....i've seen mostly A2 chips on the net....mine is most likely a real nv47, as it is rom strapped from device ID "0090" to "0091"....0090 was the NV47, i think, and "0091" is the string to name is 7800GTX.


pulled from rivatuner:

$ffffffffff ---------------------------------------------------
$ffffffffff Display adapter information
$ffffffffff ---------------------------------------------------
$0000000000 Description : NVIDIA GeForce 7800 GTX
$0000000001 Vendor ID : 10de (NVIDIA)
$0000000002 Device ID : 0091
$0000000003 Location : bus 1, device 0, function 0
$0000000004 Bus type : PCIE
$000000000f PCIE link width : 16x supported, 16x selected
$0000000009 Base address 0 : d0000000 (memory range)
$000000000a Base address 1 : c0000000 (memory range)
$000000000b Base address 2 : none
$000000000c Base address 3 : d1000000 (memory range)
$000000000d Base address 4 : none
$000000000e Base address 5 : 0000a000 (I/O range)
$ffffffffff ---------------------------------------------------
$ffffffffff NVIDIA specific display adapter information
$ffffffffff ---------------------------------------------------
$0100000000 Graphics core : NV47 revision A1 (24x1,8vp)
$0100000001 Hardwired ID : 0090 (ROM strapped to 0091)
$0100000002 Memory bus : 256-bit
$0100000003 Memory type : DDR3 (RAM configuration 07)
$0100000004 Memory amount : 262144KB
$0100000100 Core domain 0 : 275.400MHz
$0100000101 Core domain 1 : 275.400MHz
$0100000102 Core domain 2 : 275.400MHz
$0100000006 Memory clock : 326.250MHz (652.500MHz effective)
$0100000007 Reference clock : 27.000MHz
$ffffffffff ---------------------------------------------------
$ffffffffff NVIDIA VGA BIOS information
$ffffffffff ---------------------------------------------------
$1100000000 Title : ASUS EN7800GTX VGA BIOS Version 5.70.02.11.AS01
$1100000002 Version : 5.70.02.11
$1100000100 BIT version : 1.00
$1100000200 Core clock : 275MHz
$1100000201 Memory clock : 1200MHz
$1100010000 Perf. level 0 : 275MHz/600MHz/1.20V/100%
$1100010001 Perf. level 1 : 415(+35)MHz/600MHz/1.40V/100%
$1100010002 Perf. level 2 : 430(+40)MHz/600MHz/1.40V/100%
$1100020000 VID bitmask : 00000011b
$1100020100 Voltage level 0 : 1.20V, VID 00000000b
$1100020101 Voltage level 1 : 1.30V, VID 00000010b
$1100020102 Voltage level 2 : 1.40V, VID 00000001b
$ffffffffff ---------------------------------------------------
$ffffffffff NVIDIA graphics processor registers
$ffffffffff ---------------------------------------------------
$020100003c CR3C : 00000070
$0200000000 NV_PMC_BOOT_0 : 047100a1
$0200001080 NV_PBUS_DEBUG_0 : 00000000
$0200001084 NV_PBUS_DEBUG_1 : 00205649
$0200001088 NV_PBUS_DEBUG_2 : 000211f0
$020000108c NV_PBUS_DEBUG_3 : 00000090
$02000010f0 NV_10F0 : 00007fff
$0200001218 NV_1218 : 00000000
$0200001540 NV_1540 : 73ffff3f
$02000015f0 NV_15F0 : 00000000
$02000015f4 NV_15F4 : 00000000
$02000015f8 NV_15F8 : 7fffffff
$0200004000 NV_4000 : 8000011c
$0200004004 NV_4004 : 01013305
$0200004008 NV_4008 : 01c01f02
$0200004010 NV_4010 : c001011f
$0200004014 NV_4014 : 011f1002
$0200004018 NV_4018 : 0005001f
$020000401c NV_401C : 0e021104
$0200004020 NV_4020 : e400001c
$0200004024 NV_4024 : 1d030a02
$020000402c NV_402C : e400001c
$0200004030 NV_4030 : 1d030a02
$0200004038 NV_4038 : e400001c
$020000403c NV_403C : 1d030a02
$0200004044 NV_4044 : e400001c
$0200004048 NV_4048 : 1d030a02
$0200004050 NV_4050 : 00000000
$020000c000 NV_C000 : 00020010
$020000c010 NV_C010 : 00000080
$020000c014 NV_C014 : 00000000
$020000c018 NV_C018 : 00000080
$020000c01c NV_C01C : 00000000
$020000c020 NV_C020 : 00000001
$020000c024 NV_C024 : 00000000
$020000c028 NV_C028 : 00000001
$020000c02c NV_C02C : 00000000
$020000c040 NV_C040 : 3c03c333
$020000c04c NV_C04C : 10101010
$0200100200 NV_PFB_CFG : 43003073
$020010020c NV_PFB_CSTATUS : 10000000
$0200100220 NV_PFB_TIMING0 : 0813211b
$0200100224 NV_PFB_TIMING1 : 0a01060a
$0200100228 NV_PFB_TIMING2 : 20270509
$02001002c0 NV_PFB_MRS : 00000202
$02001002c4 NV_PFB_EMRS : 00100088
$0200100474 NV_100474 : 11110004
$0200101000 NV_PEXTDEV_BOOT_0 : a34ede9f
$0200101004 NV_PEXTDEV_BOOT_0_AND : 7e2fce3f
$0200101008 NV_PEXTDEV_BOOT_0_OR : 01401080
$020010100c NV_PEXTDEV_BOOT_1 : 80008490
$0200101010 NV_PEXTDEV_BOOT_1_AND : 7fffffff
$0200101014 NV_PEXTDEV_BOOT_1_OR : 00000000
$020060081c NV_PCRTC_GPIO_EXT : 62600622
$0200680500 NV_PRAMDAC_NVPLL_COEFF : 00000000
$0200680570 NV_PRAMDAC_NVPLL_COEFF_1 : 00000000
$0200680504 NV_PRAMDAC_MPLL_COEFF : 00000000
$0200680574 NV_PRAMDAC_MPLL_COEFF_1 : 00000000
$020068050c NV_PRAMDAC_PLL_COEFF_SELECT : 00000204

i dunno...but it seems pretty obvious to me that when they annouced they cancelled the NV47, that info came from TSMC....TSMC wasn't making the chip anymore, so the rumour proved true. But now it has come to light that nVidia simply moved the production to IBM fabs, either because of the yeild issues that TSMC has had as of late, or just simply to try to confuse ATI's insider @ TSMC....either way, there is more than meets the eye in the G70 story.



So, the G70...it's still out there...it's the 7800U. 32pp, 10vs. I bet IBM are producing those chips right now....7800U was cancelled...lol...like the NV47!

[Tim]

Great post Dil, we already knew it from EOCF, but it's good to see responses here..... If nVidia has something (as something is really, I mean really fishy here) they will no doubt release it after the launch of the X1800 series....

So if true, this thread becomes part of GFX history....and the guy who "figured it out".



Edit: wow there you have it

$0100000000 Graphics core : NV47 revision A1 (24x1,8vp)

7800U was cancelled...lol...like the NV47

Exactly

[DilTech]

I always knew my attention to detail would come in handy some day...

Anyway, thanks for the compliments Tim, you know I did ALOT of digging to piece this all together, and of course, you got to watch me do it all on EOCF.

Either way, I thought for sure Mesce was going to close the thread on me.

[cadaveca]

um, digitlife produced that article in june, and this beyond3d article claimed tha same in june as well:

None the less, it is always an inevitability that new products will eventually become available and NVIDIA have tried to keep people guessing a little by the change of internal codenames. Initially indications were that NV47 would be NVIDIA's next high end chip update, with suggestions of it increasing the number of internal pipelines and being based on a 110nm process. Of course, before "NV47" could come to the market codenames of G70 and G80 surfaced which possibly threw people off the scent

http://www.beyond3d.com/previews/nvidia/g70/


This is not new news, but the proof and format is new. Well done Diltech! i had thought something was fishy when i got rivatuner yesterday....guess we can confirm it now.

[[XC] leviathan18]

its not cancelled is just delayed why affect 7800gtx sales when they are making money with them and there is not real competition until next month?

[DilTech]

Nowhere in there does it say the 7800gtx is the NV47, just that codenames of G70/G80 surfaced before the NV47 could be released. It should also be noted that NVidia stated the NV47 was cancelled completely, and therefore that makes the fact that it turns out the 7800gtx IS the NV47 news, and since no site has posted the truth, that makes this New News... Othersites only posted suspicions, I posted everything with proof.

[freecableguy]

Delayed or postponed? That's the question. Big difference.

[[XC] leviathan18]

is quite the same

delayed the launch date until there is a better card from ati or postponed until the same thing

if there is any difference tell me xD

[DilTech]

My theory is simple...

ATi announce major delays, NVidia cease opportunity by launching the NV47, to make back R&D money spent. ATi still don't have card ready, but should soon, NVidia announces G71/7800ultra cancelled(to sell cards to people who *were* going to wait for it). ATi releases R520 and beats NV47, NVidia releases G71(the REAL G70!) and beats R520.

This leaves ATi in quite a bind, as it gives them 2 options, and one of which they cannot afford to do.
1.) They can take a HUGE 7 - 8 figure loss on the R520 by releasing the R580 and killing their own R520 sales, not to mention have to lower the price on the R520.
2.) take the loss of the performance crown and still sell enough cards to break even to their fanboys, as we all know fanboys don't care who's faster as long as the card is made by their favorite company...

This is my theory on the situation anyway.