AMD Vega 10, Vega 11, Vega 12 and Vega 20 Rumors and Discussion

https://videocardz.com/69760/engineering-sample-update

Vega Engineering samples caught on public benchmarks so far:

oe8bpf4.png
 
1600/945 for the FE is great news as the enthusiast version should be even higher.
 
687F:C1 being 1500Mhz doesn't bode well for Vega, it was barely ahead of 1080 in DOOM when AMD showed it off and its 3dmark scores were around 1070 level where its clock was judged to be 1200Mhz. If it's >450mm2 die size, it could be a pretty big fail at that clockspeed.
 
gamervivek said:
687F:C1 being 1500Mhz doesn't bode well for Vega, it was barely ahead of 1080 in DOOM when AMD showed it off and its 3dmark scores were around 1070 level where its clock was judged to be 1200Mhz. If it's >450mm2 die size, it could be a pretty big fail at that clockspeed.
Click to expand...
Keep in mind that those are scores caught on benchmarks so far. All but the Frontier entries are quite old.
 
Malo said:
Keep in mind that those are scores caught on benchmarks so far. All but the Frontier entries are quite old.
Click to expand...
Not only that but they may very well have been leaks from the iMac Pro where thermals would have been limited from an all in one/mobile platform.
 
Benchmarks for the iMac Pro already? Isn't the release date in the Dec timeframe?
 
pharma said:
Benchmarks for the iMac Pro already? Isn't the release date in the Dec timeframe?
Click to expand...
They were sitting on tables at WWDC running something, and all it takes is for an engineer, marketing dude, or partner to run a test. They would have had access to the engineering samples as well, prior to building an all in one device.

Checked the leaks and they all say Windows though. Clocks don't have to be final and the engineering samples are probably all the same chip.
 
gamervivek said:
687F:C1 being 1500Mhz doesn't bode well for Vega, it was barely ahead of 1080 in DOOM when AMD showed it off and its 3dmark scores were around 1070 level where its clock was judged to be 1200Mhz. If it's >450mm2 die size, it could be a pretty big fail at that clockspeed.
Click to expand...

im surprised you have not take the 1000mhz entry as reference ..

iMacmatician said:
There is a new Mac Pro planned but it won't be released this year.
Click to expand...

I have not say they will not do one, maybe a dual gpu, dual cpu setup, but honestly, i dont think it is a market who will evolve for them and attract new client base, and they know it. In fact this Imac Pro is surely more attractive for their pro client as "desktop / workstation " than a mac Pro.
 
Anarchist4000 said:
They were sitting on tables at WWDC running something, and all it takes is for an engineer, marketing dude, or partner to run a test. They would have had access to the engineering samples as well, prior to building an all in one device. .
Click to expand...
The sole iMac Pro prototype was being guarded at all times. No one was allowed to touch it, and if you somehow did, you'd be quickly noticed and, presumably, the large man nearby would remove you.
 
Ryan Smith said:
The sole iMac Pro prototype was being guarded at all times. No one was allowed to touch it, and if you somehow did, you'd be quickly noticed and, presumably, the large man nearby would remove you.
Click to expand...

The large man nearby sounds awfully convenient, I want one for my system too!

I doubt the leaked vega benchmarks came from an imac to be honest, which are the more recent ones? I find it hard to believe Vega SKUs will span a 1200-1600mhz range. Probably just misreported clocks, maybe its just reading the boost clock

~1ghz on 8Hi stacks of HBM2? How likely is that?
 
First, enginner or tester from Apple will never let an internet connection up when running testing and update the result on an database, this have never happend before .. Apple dont like unconrtrolled leak... at contrario of entrerprise or whoever got engineer test sample ( for software test etc )
 
ieldra said:
The large man nearby sounds awfully convenient, I want one for my system too!

I doubt the leaked vega benchmarks came from an imac to be honest, which are the more recent ones? I find it hard to believe Vega SKUs will span a 1200-1600mhz range. Probably just misreported clocks, maybe its just reading the boost clock

~1ghz on 8Hi stacks of HBM2? How likely is that?
Click to expand...


thoses as for most engineer samples wiith only code names .. with different clocks, meaning testing board send to early softwares tests in general.( in different countrry as China where leaks are more normal )...,,, But the FE edition dont have a codename, the complete name, meaning normally production gpus ( or end bios ).. or near production gpu's .....

If they report a 1600mhz. we should expect it to be at least 1600mhz... at least for VegaRX.. thoses numers are too much "round" .. no 1624mhz, 1224mhz for be "oc test" or clocking from self users from the board they have got .
 
Malo said:
Keep in mind that those are scores caught on benchmarks so far. All but the Frontier entries are quite old.
Click to expand...

The DOOM demo was way back in December but the Firestrike/Time Spy leaks were in April. The Compubench scores are even more recent.

lanek said:
im surprised you have not take the 1000mhz entry as reference ..



I have not say they will not do one, maybe a dual gpu, dual cpu setup, but honestly, i dont think it is a market who will evolve for them and attract new client base, and they know it. In fact this Imac Pro is surely more attractive for their pro client as "desktop / workstation " than a mac Pro.
Click to expand...

I'm not sure what you mean? Why I didn't take 1000Mhz for the 687F sample or that it's reference clock was reported as 1000Mhz?
 
ieldra said:
I doubt the leaked vega benchmarks came from an imac to be honest, which are the more recent ones? I find it hard to believe Vega SKUs will span a 1200-1600mhz range. Probably just misreported clocks, maybe its just reading the boost clock
Click to expand...

1200Mhz could be a particular workstation SKU (or mobile, but I'd expect those to come later) designed for a restrictive TDP. See for example the WX 5100, Polaris 10 cut-down and down-clocked for a 75W TDP.
 
So liquid and air cooled versions have the same clocks? The lack of distinction between the two suggests the air cooled will throttle more, while the liquid will sustain the rated 1600MHz boost, albeit at a hefty power consumption cost.
 
You must log in or register to reply here.
Back
Top