• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Poll: ** The AMD VEGA Thread **

On or off the hype train?

  • (off) Train has derailed

    Votes: 207 39.2%
  • (on) Overcrowding, standing room only

    Votes: 100 18.9%
  • (never ever got on) Chinese escalator

    Votes: 221 41.9%

  • Total voters
    528
Status
Not open for further replies.
Seems an elaborate magic trick AMD must be pulling off at this point where they are going to pull a freight train out of a top hat and we'll discover how they do more with less and it'll all make sense. HBM2 will be greater in the next iteration, seemingly true of every other feature but hopefully not completely the case
Instead of raw bandwidth Vega must be about using what its got very smartly, if true thats possibly too subtle to tell pre release.





4PxDFfB.gif

giphy.gif
 
Yh it doesn't make sense...they either ballsed up the core somehow or the driver team are struggling to unlock it.

It is all about power efficiency. They simply can't push the architecture they have anymore without pushing over a 300w tdp. I think that is exactly why it has the same amount of cores, despite being die shrunk (compared with the Fury X). Also i believe Pascal sacrificed some IPC for higher clocks compared with Maxwell so the same could be seen here.

The writing was on the wall with Polaris. The RX580 pulls more watts than a 1080, yet the 1080 is about 60-70% faster in games on average.

I'll be honest i was hoping they might have been able to improve the power efficiency problem a bit with Vega but they simply have not made good enough gains in this area and that means they are limited in what they can do.
 
Tomorrow AMD will launch new 17.7.2 display driver which will introduce some changes.



Enhanced Sync
NVIDIA’s Fast Sync finally gets a competitor. With 17.7.2 driver AMD introduces an alternative to VSync called Enhanced Sync. I’m not going to explain how this works here, but what you need to know is that it will give you the advantages of VSync without the latency caused by synchronization with the display. No more tearing and no more lag.

Relive at 100 MB/s
Relive will now allow you to save the recordings at a higher bitrate. You will now be able to record at 100 Mb/s. The Relive in 17.7.2 driver will also allow you to add a webcam to the stream and boost the volume of the microphone from within the Radeon panel.

Radeon Chill now on laptops
Radeon Chill will now be enabled for mobile graphics cards, multi-GPU configurations and also external graphics cards. In case you don’t know what Radeon Chill is, it’s basically a frame limiting tool to lower power consumption and the heat.

Full changelog will be released tomorrow.
Good news that they have a Fast-Sync thingy. It is a fantastic utility that really does work and over 100 fps (my monitors top Mhz), I never see tearing or stutter.
 
The problem, as with ryzen, AMD just cant get the clock speeds without causing a power grid shutdown.
Ryzen is propably one of the best powerconsumption chips ever produced for x86. You can even use their 8/16 desktop cpu in laptop. Even if its pushed far out of what the 14nm process was meant (3.0ghz) to do.

I don't understand. If it's running at 1600mhz why wouldn't it be a massive improvement over the Fury X. Unless the architecture actually went backwards.

For Vega to achieve that higher frequenzy they needed longer pipeline which is negative for performance in itself. Maeby they tought the process would allow Vega to go higher what its able to do now and they actually lost performance for pipeline/frequenzy.
 
Also i believe Pascal sacrificed some IPC for higher clocks compared with Maxwell so the same could be seen here.

Yeah Pascal did lose a bit of IPC but not as much as Vega is losing compared to Fiji (up to around 9% from the first tests)
And the difference here in between Fiji -> Vega and Maxwell -> Pascal is that Pascal didn't really improve on an architectural level, the things that make pascal what it is were in Maxwell, in other words Maxwell was already powerful in and of itself just limited by a bigger node. But Vega is supposed to be an architectural improvement over any GCN ever made (the biggest changes that they ever made to GCN is what they said) you wouldn't think those improvements wouldn't add IPC, that's the whole point of em, to get on the same playing field as pascal.
 
I feel like for me personally it's a terrible time to be looking for a new GPU.

I'm not keen on AMD as my graphics provider and the 1000 series have over inflated prices for what is a GPU in the twilight of it's release cycle.

Vega has to be SENSATIONAL to get my money, otherwise I'm hoping for cheaper 1080's when Vega launches.

If it wasn't for the need of a new monitor it wouldn't be such a hard choice, but a monitor you keep for 5-10 years (my current one is now 9 years old) so if I go FreeSync/Gysnc if you want to keep that feature your tied in to either Nvidia or AMD GPU's for the life of the monitor.

So it is a big decision! If it was a gamble AMD vs Nvidia my money goes on Nvidia.

---

Ultimately I'm looking for a sweet spot 1400p card that will generally get me ultraish settings with out breaking the bank.
 
Last edited:
I can't see how RTG can get close with the pricing on these GPU's given how expensive HBM et all is to produce. If indeed the top GPU is a little/equal to a 1080 then it will need to 1080 money not nearly 1080ti money, if the performance leaks are true then RTG will have very slim margins on each one sold and that's without any price reductions Nvidia could have up there sleeve (if they even need them).
 
Status
Not open for further replies.
Back
Top Bottom