• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Vega Rumors

razor1 said:
Its locked for users, but the voltage changes via driver and clock gating is functioning.

The only ? is the draw stream binned rasterizer, and that I think has to do with coding via developers.
Click to expand...

I thought in that video he said the voltage stayed pretty much the same, and didn't adjust at all. no? so it wasn't adjusting on the fly.
 
Magic Hate Ball said:
I mean, you're wrong according to David Kanter's video and Anandtech's follow up article from last year?

http://www.anandtech.com/show/10536/nvidia-maxwell-tile-rasterization-analysis

http://www.realworldtech.com/tile-based-rasterization-nvidia-gpus/

If developers had to implement these changes in their games (and not in drivers) then why Mr. Kanter have to play detective well after Maxwell's release?

EDIT: From that Techreport article relating to DSBR

To help the DSBR do its thing, AMD is fundamentally altering the availability of Vega's L2 cache to the pixel engine in its shader clusters. In past AMD architectures, memory accesses for textures and pixels were non-coherent operations, requiring lots of data movement for operations like rendering to a texture and then writing that texture out to pixels later in the rendering pipeline. AMD also says this incoherency raised major synchronization and driver-programming challenges.
Click to expand...


For Maxwell it isn't programmable nor is it programmable for Pascal. You have to remember even with Maxwel their ROPS had access to L2 cache. AMD's current GCN doesn't have that with Vega they do.

And what I linked to in techreport's article is straight from what AMD has stated. No guess there dude.

DSRB is not active all the time, nor will ever be, only in limited scenarios, so yeah its programmable....

L2 cache isn't the only part Vega needs for DSBR, to make it programmable, they had do something else to it, there needs to be access to the shader units, and current shader units won't be able to do that, and this is where the primitive shaders come in. Now you won't be doing all of this through L2
 
NKD said:
I thought in that video he said the voltage stayed pretty much the same, and didn't adjust at all. no? so it wasn't adjusting on the fly.
Click to expand...


He stated it fluctuated but to keep the mhz the same he had to up the power limits. Relatively the same he states.
 
razor1 said:
For Maxwell it isn't programmable nor is it programmable for Pascal. You have to remember even with Maxwel their ROPS had access to L2 cache. AMD's current GCN doesn't have that with Vega they do.

And what I linked to in techreport's article is straight from what AMD has stated. No guess there dude.

DSRB is not active all the time, nor will ever be, only in limited scenarios, so yeah its programmable....

L2 cache isn't the only part Vega needs for DSBR, to make it programmable, they had do something else to it, there needs to be access to the shader units, and current shader units won't be able to do that, and this is where the primitive shaders come in. Now you won't be doing all of this through L2
Click to expand...

It's not developer programmable from what I'm reading. It's a driver side thing from AMD, it will likely be game specific and programmable in that sense, but not on the developers like packed math or shader primitives (a la Doom).
 
razor1 said:
He stated it fluctuated but to keep the mhz the same he had to up the power limits. Relatively the same he states.
Click to expand...

May be but I watched the video and towards the beginning he said it stayed constant at 1150 from 900 to 1600. Didn't change one bit. I am sure you can confirm that by rewatching.
 
NKD said:
May be but I watched the video and towards the beginning he said it stayed constant at 1150 from 900 to 1600. Didn't change one bit. I am sure you can confirm that by rewatching.
Click to expand...


GCN doesn't have dynamic voltage, not that I'm aware of, it does fluctuate some small amounts, but doesn't matter, its not like Pascal's boost where it has dynamic voltage with Boost 3.0 its more like Maxwell's Boost 2.0.
 
razor1 said:
GCN doesn't have dynamic voltage, not that I'm aware of, it does fluctuate some small amounts, but doesn't matter, its not like Pascal's boost where it has dynamic voltage with Boost 3.0 its more like Maxwell's Boost 2.0.
Click to expand...
Going forward it will be curious if AMD works on something similar to XFR for their GPUs to compete better with Nvidia's boost 3.0.
 
razor1 said:
GCN doesn't have dynamic voltage, not that I'm aware of, it does fluctuate some small amounts, but doesn't matter, its not like Pascal's boost where it has dynamic voltage with Boost 3.0 its more like Maxwell's Boost 2.0.
Click to expand...

Point well made. I agree.
 
razor1 said:
When selling FE at 1000 bucks a pop they better be the better ones lol.
Click to expand...

Its been said by AMD and repeated on here as well that the FE Vega will not be the best at gaming. The RX Vega will be, how much better remains to be seen. The FE is just a niche product that AMD felt needed to be served, even I am not convince their is a market for it, but they have sold so it is what it is. Far to much focus has been on this FE Vega, will just have to see how the RX Vega shakes out.
 
Gideon said:
Its been said by AMD and repeated on here as well that the FE Vega will not be the best at gaming. The RX Vega will be, how much better remains to be seen. The FE is just a niche product that AMD felt needed to be served, even I am not convince their is a market for it, but they have sold so it is what it is. Far to much focus has been on this FE Vega, will just have to see how the RX Vega shakes out.
Click to expand...

Raja Koduri – Chief Architect Radeon Technologies Group

"Consumer RX will be much better optimized for all the top gaming titles and flavors of RX Vega will actually be faster than Frontier version!"
 
Gideon said:
Its been said by AMD and repeated on here as well that the FE Vega will not be the best at gaming. The RX Vega will be, how much better remains to be seen. The FE is just a niche product that AMD felt needed to be served, even I am not convince their is a market for it, but they have sold so it is what it is. Far to much focus has been on this FE Vega, will just have to see how the RX Vega shakes out.
Click to expand...


Not talking about raw performance, talking about asic levels. Possibility of overclocking and what not.

Anycase Videocardz seems to think Vega XTX and XT are reference design only and the XL is the only one with AIB custom coolers

XTX is already released that is the FE
Little known fact: Vega XTX is not actually such a secret anymore, it’s a codename of Frontier’s GPU (Vega 10 GLXTX)
Click to expand...


Looks like the reduction in Vram stacks didn't drop the power consumption kinda surprising there. (well TDP ratings)
 
Last edited:
razor1 said:
Not talking about raw performance, talking about asic levels. Possibility of overclocking and what not.

Anycase Videocardz seems to think Vega XTX and XT are reference design only and the XL is the only one with AIB custom coolers

XTX is already released that is the FE



Looks like the reduction in Vram stakes didn't drop the power consumption kinda surprising there. (well TDP ratings)
Click to expand...

Just gives more TDP headroom for keeping the core closer to max boost I guess.

This looks to be an exact repeat of Fury X / Fury if the full maxed out chip is only reference water cooled again.
 
razor1 said:
Not talking about raw performance, talking about asic levels. Possibility of overclocking and what not.

Anycase Videocardz seems to think Vega XTX and XT are reference design only and the XL is the only one with AIB custom coolers

XTX is already released that is the FE



Looks like the reduction in Vram stakes didn't drop the power consumption kinda surprising there. (well TDP ratings)
Click to expand...

That doesn't make sense, as from what I have seen, the XL will be the lowest Vega card with maybe only 4 GB HBM2 memory. The XTX and XT look to be the performance options.
 
Maddness said:
That doesn't make sense, as from what I have seen, the XL will be the lowest Vega card with maybe only 4 GB HBM2 memory. The XTX and XT look to be the performance options.
Click to expand...


well most likely with the less vram that will give more power options to the core itself.
 
razor1 said:
Not talking about raw performance, talking about asic levels. Possibility of overclocking and what not.

Anycase Videocardz seems to think Vega XTX and XT are reference design only and the XL is the only one with AIB custom coolers

XTX is already released that is the FE



Looks like the reduction in Vram stakes didn't drop the power consumption kinda surprising there. (well TDP ratings)
Click to expand...
It says the XTX has an AIO cooler, though, and the watercooled FE doesn't come out until later this year.

VideoCardz is part of the rumor uroboros with WCCFTech. Remember that VideoCardz also "confirmed with their sources" that the 1080 Ti would be announced at CES this year.
 
Armenius said:
It says the XTX has an AIO cooler, though, and the watercooled FE doesn't come out until later this year.

VideoCardz is part of the rumor uroboros with WCCFTech. Remember that VideoCardz also "confirmed with their sources" that the 1080 Ti would be announced at CES this year.
Click to expand...


Well these rumors come from 3dcenter just reposted by videocardz.

https://www.3dcenter.org/news/erste-informationen-zu-den-einzelnen-modellen-von-amds-radeon-rx-vega

Apparently the source is directly from an AIB partner. This matches up with Apple versions too.
 
razor1 said:
Not talking about raw performance, talking about asic levels. Possibility of overclocking and what not.

Anycase Videocardz seems to think Vega XTX and XT are reference design only and the XL is the only one with AIB custom coolers

XTX is already released that is the FE



Looks like the reduction in Vram stacks didn't drop the power consumption kinda surprising there. (well TDP ratings)
Click to expand...


Buildzoid just finished figuring out most of the voltage control on the FE, the core is entirely supplied by the 2x 8 pins the HBM and other minor stuff are ran off the PCI-e bus power so not surprising TDP wouldn't change much. what is interesting is the HBM is basically maxed out on voltage already. it seems like they are running everything neary to the max on these to get every last bit of perf out of them.

here is more info and pics from BZ himself. https://imgur.com/a/ImPz1
 
Frito11 said:
Buildzoid just finished figuring out most of the voltage control on the FE, the core is entirely supplied by the 2x 8 pins the HBM and other minor stuff are ran off the PCI-e bus power so not surprising TDP wouldn't change much. what is interesting is the HBM is basically maxed out on voltage already. it seems like they are running everything neary to the max on these to get every last bit of perf out of them.

here is more info and pics from BZ himself. https://imgur.com/a/ImPz1
Click to expand...


4 slots to keep this thing without throttling wow!
 
Ieldra said:
Essentially, anarchist, we went from 13.1tflops at 300W with headroom, as per your claims, to the impossibility of such throughput at stock settings and a 400+w requirement it actually meet it
Click to expand...
We haven't established anything yet as there aren't any relevant benchmarks. Just worst case examples that can't be used to draw any conclusions. Card forced to higher than necessary voltages with power saving features disabled uses more power. Real shocker there. I'm still liking my claims.
 
Frito11 said:
Buildzoid just finished figuring out most of the voltage control on the FE, the core is entirely supplied by the 2x 8 pins the HBM and other minor stuff are ran off the PCI-e bus power so not surprising TDP wouldn't change much. what is interesting is the HBM is basically maxed out on voltage already. it seems like they are running everything neary to the max on these to get every last bit of perf out of them.

here is more info and pics from BZ himself. https://imgur.com/a/ImPz1
Click to expand...
That likely won't result in getting the most performance out of them. Thermal throttling will occur regardless of voltages. I'm still shocked those 8-Hi stacks aren't melting already.
 
Radeon Vega Fermi edition XTX, XT, XL. The worst part is i'll probably end up buying one at some point due to my monitor, lol!
 
I suppose those are just GPU codenames, like Fiji XT and so on.
Do people expect to see "AMD RX Vega XTX" on the box? Because you won't.

Then again I also didn't believe Nvidia was dumb enough to release a GPU called "GTX 1080" but they did...
 
I want the XXX huge dick edition that takes up five slots. Give me a man's blower.

So 10-15% more on clocks with crazy cooling? Should at least be 1080 perf now.. that's reasonable until Volta hits. At least it can run games on high/ultra on a 21:9.
 
Dayaks said:
I want the XXX huge dick edition that takes up five slots. Give me a man's blower.

So 10-15% more on clocks with crazy cooling? Should at least be 1080 perf now.. that's reasonable until Volta hits. At least it can run games on high/ultra on a 21:9.
Click to expand...

Pun intended? :censored: Left overs from the robot toys thread lol.
 
razor1 said:
well most likely with the less vram that will give more power options to the core itself.
Click to expand...

That still doesn't make sense. The card with the lowest memory count having the best performance. I don't buy that for a second.
 
I am suspecting that HBM2 this time is using more power as well. I wonder if they can check how much power alone the memory is using. I looks like it is really pushed to max and AMD really got screwed with the HBM2 issues.
 
asic wise 300 watts just for the GPU for the water cooled and 220 watts for the air cooled. so 65 to 85 watts for the rest of the board. Most of the left board power should be eaten up by the vram.
 
Maddness said:
That still doesn't make sense. The card with the lowest memory count having the best performance. I don't buy that for a second.
Click to expand...


Don't know man, just wondering how they are going to get more clocks out of the RX version if its already pretty much power limited by the pci-e specs.
 
Maddness said:
That still doesn't make sense. The card with the lowest memory count having the best performance. I don't buy that for a second.
Click to expand...
Saw elsewhere they mentioned RX may use the higher bandwidth HBM2 because it has only 8Gb and the 16Gb needed lower for stability.
 
You must log in or register to reply here.
Back
Top