RX Vega Rumored to Hash 70+ Mh/s

[Metroid]

Exactly, however, they are spouting nonsense.

[vuli]

ajaj, I am afraid you people are dreaming a little too much. At the end we all are going to be disapointed.

[shimpa]

so why is Vega FE not performing the same or at least in the same ballpark? Same GPU (well as far as mining goes it's identical), same memory (only more of it on FE) so I fail to see how the RX Vega can double or TRIPLE the hashrate. Driver optimization, okay do it with FE it shouldn't really matter.

[Metroid]

so why is Vega FE not performing the same or at least in the same ballpark? Same GPU (well as far as mining goes it's identical), same memory (only more of it on FE) so I fail to see how the RX Vega can double or TRIPLE the hashrate. Driver optimization, okay do it with FE it shouldn't really matter.

VEGA Frontier is not a gaming card. But I do agree concerning the hashrate jump to 70mhs to be too much even with somethings enabled on the gaming version.

[jstefanop]

so why is Vega FE not performing the same or at least in the same ballpark? Same GPU (well as far as mining goes it's identical), same memory (only more of it on FE) so I fail to see how the RX Vega can double or TRIPLE the hashrate. Driver optimization, okay do it with FE it shouldn't really matter.

VEGA Frontier is not a gaming card. But I do agree concerning the hashrate jump to 70mhs to be too much even with somethings enabled on the gaming version.

Vega FE in its current form has its HBM2 working ~ 400 GB/s...about 40GB/s of that discrepancy is actually from the upper stacks on the 16GB HBM2 (FE has two stacks of two 4GB dies on top of each other..the top two dies actually run at lower timings due to hardware limitations...ie top stacks run slower since electrical signals have to go through additional vias to reach top). The other 40GB is due to driver optimizations that AMD is working on right now.

So out of the box we can expect RX Vega to do 440GB/s with its single stacked two dies (8GB total HBM2), plus near 480GB/s if they fixed the driver issues by launch. Thats a 20% bump from the stock ~33-35MH = ~ 41MH. Memory can be overclocked at least 20% from my tests, so thats brings it up to ~50 MH/s. Throw in another 10% with timing optimizations and were looking at about 55MH top eth speed for vega.

Disclamer I DO have a Vega FE and and working with AMD devs to address these issues and was at the siggraph event and talked to Raja. AMD is finally in a corner where they can't ignore the mining market anymore, and I'm doing the best I can do address the mining markets concerns to AMD. Expect more good stuff from them in the future

[QuintLeo]

and the time will come as the grave for 1070/1080ti , huhahahaha bye2 nvidia miner

 Only reason the 1070 got popular for ETH was that pricing got crazy on the RX 470/480/570/580 cards and made the "slightly higher" hashrate of the 1070 on ETH competative on a $/hash basis.

 I've yet to see ANY benchmarks for a Vega on ZEC - but it's going to have to top the 1070/1080/1080ti a LOT on hashrate to get close on efficiency.



 The Titan Xp "major performance upgrade" was pretty much only on ONE specific application, and to a lesser degree some very similar applications.
 It wasn't a "improve performance 3x on everything" type upgrade, but more a "fix a problem this specific application was having" type fix.


 Gamers are NOT waiting on GTX 1080ti - pricing and availability has been pretty much UNTOUCHED on that particular model.
 It's the 1070, 1060, and to a VERY VERY MINOR degree the 1080 that have been affected by mining - and I'm pretty sure the 1080 effects were INDIRECT from folks that had originally planned to get a 1070 but go for a 1080 instead due the price being almost the same and much greater availability.



 Ryzen isn't "destroying" Intel CPUs - but it IS being competative enough to force Intel to adapt to having to COMPETE again, after several years of unquestioned supremecy.
 Not the first time by a long shot for Intel - they lost the performance crown ENTIRELY on 3 previous occasions (Athlon 64, Athlon Thunderbird, Z80) and AMD also challenged them very closely in the days of the K5 vs the Pentium (K5 was SUPERIOR on integer operations per clock and close enough on clock to match or beat overall integer performance, but lost badly on floating point).

[Metroid]

Disclamer I DO have a Vega FE and and working with AMD devs to address these issues and was at the siggraph event and talked to Raja. AMD is finally in a corner where they can't ignore the mining market anymore, and I'm doing the best I can do address the mining markets concerns to AMD. Expect more good stuff from them in the future

Good to know, keep up the good work.

[philipma1957]

so why is Vega FE not performing the same or at least in the same ballpark? Same GPU (well as far as mining goes it's identical), same memory (only more of it on FE) so I fail to see how the RX Vega can double or TRIPLE the hashrate. Driver optimization, okay do it with FE it shouldn't really matter.

VEGA Frontier is not a gaming card. But I do agree concerning the hashrate jump to 70mhs to be too much even with somethings enabled on the gaming version.

Vega FE in its current form has its HBM2 working ~ 400 GB/s...about 40GB/s of that discrepancy is actually from the upper stacks on the 16GB HBM2 (FE has two stacks of two 4GB dies on top of each other..the top two dies actually run at lower timings due to hardware limitations...ie top stacks run slower since electrical signals have to go through additional vias to reach top). The other 40GB is due to driver optimizations that AMD is working on right now.

So out of the box we can expect RX Vega to do 440GB/s with its single stacked two dies (8GB total HBM2), plus near 480GB/s if they fixed the driver issues by launch. Thats a 20% bump from the stock ~33-35MH = ~ 41MH. Memory can be overclocked at least 20% from my tests, so thats brings it up to ~50 MH/s. Throw in another 10% with timing optimizations and were looking at about 55MH top eth speed for vega.

Disclamer I DO have a Vega FE and and working with AMD devs to address these issues and was at the siggraph event and talked to Raja. AMD is finally in a corner where they can't ignore the mining market anymore, and I'm doing the best I can do address the mining markets concerns to AMD. Expect more good stuff from them in the future

50-55 would be okay for eth.

what about sols for zec?

[SCSI2]

50-55 would be okay for eth.

what about sols for zec?

And what about Sol/Watt?

[jsanzsp]

so why is Vega FE not performing the same or at least in the same ballpark? Same GPU (well as far as mining goes it's identical), same memory (only more of it on FE) so I fail to see how the RX Vega can double or TRIPLE the hashrate. Driver optimization, okay do it with FE it shouldn't really matter.

VEGA Frontier is not a gaming card. But I do agree concerning the hashrate jump to 70mhs to be too much even with somethings enabled on the gaming version.

Vega FE in its current form has its HBM2 working ~ 400 GB/s...about 40GB/s of that discrepancy is actually from the upper stacks on the 16GB HBM2 (FE has two stacks of two 4GB dies on top of each other..the top two dies actually run at lower timings due to hardware limitations...ie top stacks run slower since electrical signals have to go through additional vias to reach top). The other 40GB is due to driver optimizations that AMD is working on right now.

So out of the box we can expect RX Vega to do 440GB/s with its single stacked two dies (8GB total HBM2), plus near 480GB/s if they fixed the driver issues by launch. Thats a 20% bump from the stock ~33-35MH = ~ 41MH. Memory can be overclocked at least 20% from my tests, so thats brings it up to ~50 MH/s. Throw in another 10% with timing optimizations and were looking at about 55MH top eth speed for vega.

Disclamer I DO have a Vega FE and and working with AMD devs to address these issues and was at the siggraph event and talked to Raja. AMD is finally in a corner where they can't ignore the mining market anymore, and I'm doing the best I can do address the mining markets concerns to AMD. Expect more good stuff from them in the future

HBM memory Do not leave, change stock strap timing

Let's not be so optimistic, Vega's hashrate will not reach 45 mh/s never

Fury is trash in eth
Vega will be the same

[fmz89]

so why is Vega FE not performing the same or at least in the same ballpark? Same GPU (well as far as mining goes it's identical), same memory (only more of it on FE) so I fail to see how the RX Vega can double or TRIPLE the hashrate. Driver optimization, okay do it with FE it shouldn't really matter.

VEGA Frontier is not a gaming card. But I do agree concerning the hashrate jump to 70mhs to be too much even with somethings enabled on the gaming version.

Vega FE in its current form has its HBM2 working ~ 400 GB/s...about 40GB/s of that discrepancy is actually from the upper stacks on the 16GB HBM2 (FE has two stacks of two 4GB dies on top of each other..the top two dies actually run at lower timings due to hardware limitations...ie top stacks run slower since electrical signals have to go through additional vias to reach top). The other 40GB is due to driver optimizations that AMD is working on right now.

So out of the box we can expect RX Vega to do 440GB/s with its single stacked two dies (8GB total HBM2), plus near 480GB/s if they fixed the driver issues by launch. Thats a 20% bump from the stock ~33-35MH = ~ 41MH. Memory can be overclocked at least 20% from my tests, so thats brings it up to ~50 MH/s. Throw in another 10% with timing optimizations and were looking at about 55MH top eth speed for vega.

Disclamer I DO have a Vega FE and and working with AMD devs to address these issues and was at the siggraph event and talked to Raja. AMD is finally in a corner where they can't ignore the mining market anymore, and I'm doing the best I can do address the mining markets concerns to AMD. Expect more good stuff from them in the future

HBM memory Do not leave, change stock strap timing

Let's not be so optimistic, Vega's hashrate will not reach 45 mh/s never

Fury is trash in eth
Vega will be the same

i think you already stock so much rx 400/500 or gtx 1000 series, worried about less earning when vega hit market, just sell all your card now, dont cry later
claymore tested vega fe at 41mhs with his miner v9.8, no doubt if rx vega can reach 60++mhs

[Zeix]

Just do some calculations. Assuming you can undervolt them to 200W.

50 mhs / 200W = 0,25
55 mhs / 200W = 0,275
60 mhs / 200W = 0,3

GTX 1060:
23,5 mhs / 80W = 0,29375

GTX 1050 Ti:
15 mhs / 50W = 0,3


At 200W, they need AT LEAST 60 mhs to be even on par with a GTX 1050 Ti.

[kuttysark420]

yep it uses much much more power.. needs 2 be at least 60 mhs MINIMUM

[P00P135]

Gamers Nexus Confirmed but not tested RX Vega to hash 70Mh/s.  I trust this source a lot more than the previous, but again still untested.

https://youtu.be/qrnkbmabK4U?t=55

[DragonSlayer]

Some more rumors out in the wild.

https://www.overclock3d.net/news/gpu_displays/retail_staff_reports_high_levels_of_mining_performance_on_amd_s_rx_vega/1

[BellicoseBenny]

Same lies

Please people do not reward AMD for lying to us

[totko]

If i know AMD ,these cars theoretically could do it yes, but drivers are bad and there is no miner   support so i don't think its possible. But nice daydreaming! :-)

[QuintLeo]

Some more rumors out in the wild.

https://www.overclock3d.net/news/gpu_displays/retail_staff_reports_high_levels_of_mining_performance_on_amd_s_rx_vega/1

 SAME rumor just a different site quoting Gibbo.

[QuintLeo]

Gamers Nexus Confirmed but not tested RX Vega to hash 70Mh/s.  I trust this source a lot more than the previous, but again still untested.

https://youtu.be/qrnkbmabK4U?t=55

 "haven't tested it" right from the video - so no real confirmation.

[MLN84]

The reviews are out, and shops are selling (in Germany). Anyone got his hands on one yet?

Edit: golem.de obviously tried mining with stock settings for the RX Vega 64, getting 31 MHS.

I'm curious what will be possible with tweaks, custom bios, etc.