Wednesday, July 31st 2019

NVIDIA RTX 2060 Super and RTX 2070 Super Chips Come in Three Variants Each. Flashing Possible?

While working on GPU-Z support for NVIDIA's new GeForce RTX Super cards, I noticed something curious. Each of the RTX 2060 Super and RTX 2070 Super is listed with three independent device IDs in the driver: 1F06, 1F42, 1F47 for the former and 1E84, 1EC2, 1EC7 for the latter. GeForce RTX 2080 Super on the other hand, like nearly every other NVIDIA SKU, uses only a single device ID (1E81). The PCI device ID uniquely identifies every GPU model, so the OS and driver can figure out what kind of device it is, what driver to use, and how to talk to it. I reached out to NVIDIA, for clarification, and never heard back from them besides an "interesting, I'll check internally" comment.

With no official word, I took a closer look at the actual values and remembered our NVIDIA segregates Turing GPUs article, that was part of the launch coverage for the initial GeForce RTX unveil. In that article, we revealed that NVIDIA is creating two models for each GPU, that are identical in every regard, except for name and price. If board partners want to build a factory-overclocked card, they have to buy the -A variant of the GPU, because only that is allowed to be used with an out of the box overclock. Manual overclocking by the users works exactly the same on both units.
For newer releases, NVIDIA abandoned that concept again, selling just a single GPU for each SKU. But then why three device IDs? Multiple variants of GeForce RTX Super cards are definitely out there, as we have received e-mails and seen comments on our forums, from buyers wondering why their GPU "is different," how it affects performance, or if it could even be a fake card.

In the above list, I reordered and grouped the entries in the screenshot to help guide the eye. While the first SUPER entry in each group has a device ID matching the NVIDIA Founders Edition, provided to us by NVIDIA for our reviews, there are two more entries for each SKU. If you take a close look at the device IDs, you'll realize that each one just differs by a value of 40 Hex from the another models (that's what the red arrows indicate). Why would NVIDIA create those SKUs, based on existing models? It seems that in order to protect existing inventory in warehouses of their board partners, NVIDIA has created a method to turn existing RTX 2070 non-Super cards into RTX 2060 Super; and RTX 2080 non-Super to RTX 2070 Super, respectively. This also explains why there's two new IDs for each card: both non-A and -A GPUs can be converted.

We imagine this is accomplished by simply re-configuring the cards to disable some components, but the method employed could be a bit more sophisticated than simply changing the video-BIOS and slapping on a new sticker. The reason why 40 Hex was chosen as a differentiator between both, is that it represents just a single bit difference in the device ID. Such a bit change is the smallest possible change, and we suspect it is implemented through a small resistor that gets soldered onto the PCB (or removed), which the GPU probes during its internal startup, and adjusts its reported device ID accordingly. It's for this reason that it may not be able to flash an RTX 2060 Super back to an RTX 2070 provided the PCBs and ASIC match; or flashing an RTX 2070 Super to an RTX 2080.

Now of course you're wondering what reasons board partners would have to change existing, perfectly fine cards sitting in warehouses, ready to go, into slower (= cheaper) variants. Let's look at the financials quickly. RTX 2080 currently sells for $640, flashing that into RTX 2070 SUPER, selling for $520 makes little economic sense. Things don't look that much better for the $450 RTX 2070 to $400 RTX 2060 SUPER transition. Even if a board partner is having a hard time selling RTX 2080 for $630, they could definitely sell them for $600, and still make a better profit than turning it into a $520 RTX 2070 SUPER. My guess is that NVIDIA wants to increase available inventory for SUPER, reduce inventory of non-SUPER at the same time, and avoid a price fight between non-SUPER end-of-life models and brand-new SUPER cards. If that is the case, then maybe the company is giving its partners some sort of compensation, either as cash, or as rebate for future GPU purchases.

The only other explanation that I can imagine is that if demand for RTX SUPER greatly exceeds that for the older models, then partners could be tempted to convert just enough inventory to SUPER, to ensure that they don't lose sales to other AICs. People who want RTX SUPER now might look at options beyond their established hardware vendor, which could result in lost sales. This is more of a possibility nowadays, with differences between custom-design cards being minimal and coming down mostly to cosmetic differences and cooler / thermal settings. Flashing down small batches would ensure customers are kept happy, while vendors are able to reduce existing warehouse stocks a little bit quicker-essentially a balancing tool.

We're not sure exactly how the board modification is performed, whether solder rework and BIOS flash are done at the AIC factories, or whether NVIDIA does it in-house to protect their methods. We suspect this to be the case, because both the GPU and NVFlash BIOS flashing software have several layers of protection against GPU cross-flashing, and NVIDIA might not want to give its board partners the keys to the kingdom. On the other hand, if these cards can be flashed back (even if using an external flasher), and the solder mod is reasonably easy to do, then this could end up being an amazing method for people to flash their SUPER cards back to RTX 2070 / 2080 and gain some free performance in the process. These cards are full RTX 2080 / 2070 cards after all, there can be no defective units, otherwise that cards wouldn't have made it through QA in the first place.

Another interesting aspect is how SLI compatibility between cards of the same SKU, but with difference device IDs will be affected. SLI support traditionally required device IDs of two cards to match, the sub-vendor IDs can be different (i.e., you can freely pair a ZOTAC GTX 980 with an ASUS GTX 980, but you can't pair an ASUS GTX 980 with an ASUS GTX 970). We're reaching out to NVIDIA once more, to at least get feedback on that specific question.
Add your own comment

21 Comments on NVIDIA RTX 2060 Super and RTX 2070 Super Chips Come in Three Variants Each. Flashing Possible?

#1
kid41212003
I was hoping I could get my non Super 2080 to Super level :))
Posted on Reply
#2
W1zzard
kid41212003, post: 4090101, member: 55555"
I was hoping I could get my non Super 2080 to Super level :))
Unfortunately not, there is only a single RTX 2080S SKU, while there's three for 2070S / 2060S
Posted on Reply
#3
danbert2000
I had no idea they were potentially binning these cards. I got pretty much the cheapest 2070 Super that EVGA offers, the 2070 Super Black with the two fans rather than the blower. I will have to check which one I have. I'm not upset by the available overclock, I'm peaking around 1950 MHz with an OC Scanner curve, and that fits my needs just fine. If I lost 50-100 MHz by getting a cheaper card, so be it. But it would have been nice to know they were separating the chips. Perhaps there is a power limit specified for the cards differently? That could make sense if it's not like a usual card and some chips can take more power than others. I know I top out at around 235 watts as a power limiter, I could see that being up to Nvidia this time and not the surrounding board, and it would explain why a lot of the early aftermarket boards from MSI and others have had lower power limits out of the box than the "Founder's Edition," but this time there's no difference in boost clocks.

EDIT: Well, I didn't realize there was a whole other section after the break. It appears this is all about converting old cards to the new SKUs. In that case, only coverting the non-overclock GPUs to 2070 Supers should lead to a definite ASIC quality decrease. I'm still curious, since I ordered my card on July 9th. I don't know if that makes me more or less likely to have gotten a "converted" 2080. I have no interest in BIOS flashing so it doesn't really matter to me.
Posted on Reply
#5
Renald
So to be simple : manufacturers can name their old RTX 2070 into a RTX 2060 SUPER, just by changing the ID on the board, and it's fully authorized by Nvidia ?

It's not "that bad", but people should know when it's not a brand new cards.
They are coming back with rebranding, instead of assuming they priced it wrong in the first place.
Posted on Reply
#6
OneCool
A 2060s to a 2080s would be sweet :pimp:
Posted on Reply
#7
danbert2000
Renald, post: 4090182, member: 130756"
So to be simple : manufacturers can name their old RTX 2070 into a RTX 2060 SUPER, just by changing the ID on the board, and it's fully authorized by Nvidia ?

It's not "that bad", but people should know when it's not a brand new cards.
They are coming back with rebranding, instead of assuming they priced it wrong in the first place.
These GPUs haven't been used yet. They're not going to tear cards apart to convert them in my opinion. But they probably have a stack of GPUs ready to be integrated into cards that they could now switch. In either case, these cards are brand new, they just might be repurposed to serve the greater economic picture.

There's no benefit to having a 2070 Super that was disabled by Nvidia over a 2080 that became a 2070 Super.
Posted on Reply
#8
dj-electric
@W1zzard Can you tell us what happens with a Palit RTX 2060S JS to RTX 2070 Jetstream conversion? i suspect there might be something there
Posted on Reply
#9
efikkan
I'm just surprised this is news to you, don't you make that GPU-Z tool in-house?
Nvidia have been doing this for years, like:
code:

NVIDIA_DEV.1B83 = "NVIDIA GeForce GTX 1060 6GB"
NVIDIA_DEV.1B84 = "NVIDIA GeForce GTX 1060 3GB"
NVIDIA_DEV.1C02 = "NVIDIA GeForce GTX 1060 3GB"
NVIDIA_DEV.1C03 = "NVIDIA GeForce GTX 1060 6GB"
NVIDIA_DEV.1C04 = "NVIDIA GeForce GTX 1060 5GB"
NVIDIA_DEV.1C06 = "NVIDIA GeForce GTX 1060 6GB"


code:

NVIDIA_DEV.1185 = "NVIDIA GeForce GTX 660"
NVIDIA_DEV.1195 = "NVIDIA GeForce GTX 660"
NVIDIA_DEV.11C0 = "NVIDIA GeForce GTX 660"


The subject have been mentioned in the forums many times, so this by itself isn't newsworthy, but an official explanation might be.
I've just assumed it's either market segmentation (QS, OEM, founders edition, partner edition, "China edition") and/or iterative changes(steppings). But for one reason or another, Nvidia find it useful to track different editions of the product, probably to debug problems.
Posted on Reply
#11
phanbuey
the 2080ti comes in 2 variants and flashing across them is not possible using nvflash. maybe another utility can do it.
Posted on Reply
#12
W1zzard
phanbuey, post: 4090409, member: 45008"
the 2080ti comes in 2 variants and flashing across them is not possible using nvflash. maybe another utility can do it.
Both RTX 2080 Ti chips are -A and non-A from the original Turing launch. They are identical, so no point flashing them

efikkan, post: 4090280, member: 150226"
I'm just surprised this is news to you, don't you make that GPU-Z tool in-house?
Nvidia have been doing this for years, like:
code:

NVIDIA_DEV.1B83 = "NVIDIA GeForce GTX 1060 6GB"
NVIDIA_DEV.1B84 = "NVIDIA GeForce GTX 1060 3GB"
NVIDIA_DEV.1C02 = "NVIDIA GeForce GTX 1060 3GB"
NVIDIA_DEV.1C03 = "NVIDIA GeForce GTX 1060 6GB"
NVIDIA_DEV.1C04 = "NVIDIA GeForce GTX 1060 5GB"
NVIDIA_DEV.1C06 = "NVIDIA GeForce GTX 1060 6GB"


This is different. The chips in your list above are bought for that specific SKU from NVIDIA before you build the board. you're not allowed to buy a 1B84 GP104 and then use it on a GTX 1060 6 GB - only 3 GB is possible. 1C0x are the original GTX 1060 based on GP106 (not GP104 like 1B8x, which were released at a later time).

What is happening today is that completely assembled cards are flashed to variants with fewer shaders, so there is probably a way to undo that.

Think: GTX 1070 Ti cards are flashed to GTX 1070 and sold as GTX 1070, at GTX 1070 price
Posted on Reply
#13
phanbuey
It's a secret nvidia ploy to get people to brick their 2070 supers trying to flash them to 2080s, thereby increasing sales when they have to buy new replacement cards.
Posted on Reply
#14
newtekie1
Semi-Retired Folder
I suspect this might be just to allow AIBs to get Super models out the door quickly by using existing inventory of the higher end cards. They've done this in past, the GTX465 comes to mind. Some of the initial batch of GTX465 cards was flashable to GTX470. The GTX465 was similar to the Super cards in how it was launched too. It was brought out rather rushed as a response to an AMD launch just like the Super cards.
Posted on Reply
#15
Tech$sign
Cool. Great work in figuring this all out and props to possibly one of the few guys who this info leads to anything usable. If I backtrack a lil, the key is that every card/device has to be made with all of the configurations that a device can take, so it must be part of its makeup when creating so a change is now possible? By seeing these three possibilities in the device ID, leads you to understand that it may be possible to make a card take on the id of another.
Just trying to understand the thought thread that you had when figuring this out.

Again, great work......and great programmer....or should i say app developer :)
Posted on Reply
#17
phanbuey
https://pcper.com/2019/07/multiple-variants-of-rtx-2060-and-2070-super-gpus-spotted/

DStealth, post: 4091105, member: 9888"
Great article, as NVFlash 5.560 cannot be forced to cross flash, when we could expect a newer version of NVIDIA NVFlash with Board Id Mismatch Disabled like this https://www.techpowerup.com/download...atch-disabled/
:p
It would be nice to override Device ID as well. Especially for the new GPU's as the pwr cap is a huge limiting factor.
Posted on Reply
#18
John Naylor
danbert2000, post: 4090154, member: 165365"
I had no idea they were potentially binning these cards. I got pretty much the cheapest 2070 Super that EVGA offers, the 2070 Super Black with the two fans rather than the blower. I will have to check which one I have. I'm not upset by the available overclock, .....
The main differences between the Black and other AIB / FE cards:

Black uses the TU106-400-A1 instead of TU106-400A-A1
Black uses a 6+2 VRM setup, instead of 8+2
No Backplate
Takes up 2 rather than 3 slots (FE does too)
One 8 pin rather than the 8 + 6

One would think that would have a larger impact, but in the end, overclocked it's only a 8 fps difference in overclocked performance.
Posted on Reply
#19
danbert2000
John Naylor, post: 4091562, member: 156078"
The main differences between the Black and other AIB / FE cards:

Black uses the TU106-400-A1 instead of TU106-400A-A1
Black uses a 6+2 VRM setup, instead of 8+2
No Backplate
Takes up 2 rather than 3 slots (FE does too)
One 8 pin rather than the 8 + 6

One would think that would have a larger impact, but in the end, overclocked it's only a 8 fps difference in overclocked performance.
I don't think any of this applies to the 2070 Super. I have 8 + 6 on this card, and also all 2070 Supers are based on TU104 now, though clearly from the post it could be any of 3 different chips, one of which would be a harvested 2080 non-overclock version. But that GPU is still going to have pretty high clocks because it was meant for a 2080 with a higher boost clock than the 2070 non-A GPU that is getting used in 2060 Supers presumably.

There are no reviews of my card by anyone that looked at the VRM, so it could be weaker I suppose, but the TDP is up to 215 W so I doubt it's just a copy of the 2070 VRM setup. I definitely do not have a backplate. Which for my case, is actually not a downside, since it's a "sandwich" case that uses a riser cable to bend the GPU to be flat against the back of the motherboard (with the support structure of the midplane between them).

Overall I've been really happy with the 2070 Super Black. Not sure why anyone would spend more than $10-20 on a better cooler. My OC Scanner overclock has me at 1950-1980 MHz, and the 235 top end power limit is about all I could keep cool in my case anyway.

@W1zzard Are you guys going to get a 2070 Super Black in the lab? I remember your 2080 Black review and it was pretty positive, would be interested in seeing more 2070 Super reviews as most are for the same MSI and Nvidia cards.
Posted on Reply
#20
DStealth
Vipeax made 5.567 moded version. But still gives GPU Mismatch while trying to flash 2080 BIOS as devIDs are different 1E84 vs 1E87
Posted on Reply
#21
Midland Dog
OneCool, post: 4090191, member: 20830"
A 2060s to a 2080s would be sweet :pimp:
different die, tu106 is for 2060, 2060s and 2070. tu104 is for 2070s, 2080 and 2080s
Posted on Reply
Add your own comment