He says thats its dropping in GPU Usage % often as well, though sadly he doesn't show / overlay it.
AMD Vega Hardware Reviews
- Thread starter ArkeoTP
- Start date
He says thats its dropping in GPU Usage % often as well, though sadly he doesn't show / overlay it.
I haven't seen it and with the card hitting 800mhz and below so many times in these videos and reviewers saying it would have low gpu utilization gives me hope that we get more performance from the gaming edition. I'd be happy to buy one if it falls between a 1080 and ti
D
Deleted member 13524
Guest
IMHO, this kind of posts is the cue to lock the thread until RX Vega reviews are out, or at least leaked results that are significantly different to the ones we all have already.
Otherwise, in 3 weeks time what we'll have here is 10 pages of posts saying "OMG I can't believe they screwed us" over and over.
I disagree about locking the thread - not about the uselessness of the post quoted. This is the Vega Hardware Reviews thread, not the Vega RX-Only Gaming Reviews thread. So, IMHO, all testing is valid since it's a product which is on (some rare) shelves for eveyrone to buy.
D
Deleted member 13524
Guest
That's correct, then I suggest we open a thread called "99-ways I'm angry at AMD about Vega" in the general section and make that kind of "venting" posts invalid for this thread.
But to be honest, regardless of what the Vega FE is meant to be, the only thing being taken into consideration in this thread is gaming performance. Several people and websites have tested its gaming performance and it's a known value at this point. And I doubt Vega FE will get any significant driver upgrade until RX Vega is out.
Therefore, until Siggraph the discussion value of this thread is very limited, considering we already have an active thread for discussing the architecture and the latest leaks.
I'm a bit surprised that a lot of people here are drawing conclusions about VEGA based on VEGA FE gaming performance alone. VEGA RX as a gaming card can be, first of all, faster than FE (unknown amount), will for sure be a lot cheaper than FE (undercutting GTX1080 at similar perf. or matching it's price at higher perf.), might be using new stepping due to gap between launch of FE and RX and so on.
I personally would have no problem with performance and power of current VEGA FE if priced at around $400 for gamer/home user. If AMD makes enough profit on that it's their problem and they know it. AMD can't sell slower and more power hungry card for the same price as nVidia, so they have to price it accordingly.
I personally would have no problem with performance and power of current VEGA FE if priced at around $400 for gamer/home user. If AMD makes enough profit on that it's their problem and they know it. AMD can't sell slower and more power hungry card for the same price as nVidia, so they have to price it accordingly.
Vega FE is a product you can buy and so the customer deserves a working driver and product. To be honest there is, imho, no way AMD can make this look good again. If the performance does not improve much with RX Vega it is bad, but if they release a ueber driver with RX Vega, it also leaves a stale taste in the mouth, because AMD thought it would be fine to have customers of a 1000$ card work with an intentionally bad driver just for marketing purposes.
Looks like the drivers are working.
Everyone needs to pick their words more carefully to be less charged with emotion and hyperbole. Right now those sort of words make the people using them look foolish or childish with such emotional outbursts.
Yes, they could be more performant, but they are working. I dont see crashes to desktops or software refusing to run. Thus the drivers are working.
Everyone needs to pick their words more carefully to be less charged with emotion and hyperbole. Right now those sort of words make the people using them look foolish or childish with such emotional outbursts.
Yes, they could be more performant, but they are working. I dont see crashes to desktops or software refusing to run. Thus the drivers are working.
Heh remember the days that nvidia would release a product and then months later they would release much better drivers when faced with competition to raise thier performance as much as possible
D
Deleted member 13524
Guest
This just in: AMD is spending 95% of their driver devs on finding out ways to make Vega totally suck at mining and most of all to prevent it from ever stopping to suck at mining.
/s
Seriously though, AMD might be on to something here.
Despite the currently mediocre gaming performance.
Meanwhile, it's also been said that the Pro and gaming drivers are being developed by different branches of AMD's software team. Considering how there are zero performance or stability differences between the Pro and Gaming modes in games, perhaps it's a safe assumption that the "Game mode" in FE is currently just running the Pro driver while enabling wattman.
Perhaps you can blame AMD for their driver teams not having a proper gaming driver on time for Vega FE's launch, but you can't blame them for something they obviously didn't do, which is to purposely stifle their own product.
/s
Seriously though, AMD might be on to something here.
Perhaps because the card actually is performing as intended to whom the card is concerned: game devs, content creators and compute clients. SPECviewperf and FP16 compute results (except mining) (EDIT: removed compubench scores, changed to PF16) do show a sizeable price/performance advantage over the existing competition.
Despite the currently mediocre gaming performance.
The "intentionally bad driver" idea makes no sense and has been denied by AMD officials. For marketing purposes it's just as useless, because at this point even more customers who were holding out to see how Vega performs have decided to get a 1080/1080Ti instead. I'm guessing shareholders aren't particularly happy with this either. If there was a better driver for games that was stable enough, it would be public by now.
Meanwhile, it's also been said that the Pro and gaming drivers are being developed by different branches of AMD's software team. Considering how there are zero performance or stability differences between the Pro and Gaming modes in games, perhaps it's a safe assumption that the "Game mode" in FE is currently just running the Pro driver while enabling wattman.
Perhaps you can blame AMD for their driver teams not having a proper gaming driver on time for Vega FE's launch, but you can't blame them for something they obviously didn't do, which is to purposely stifle their own product.
Last edited by a moderator:
I hadn't actually seen compubench results *after* it launched and I noticed that the ones you linked were on OSX for the txp.
These are both running windows, I just went to the compare section of their website, compubench2.0, there's no mining though, and some tests are missing. No idea how valuable compubench is
Well, that depends. If there are really so many features deactivated as claimed and considering when they showed the first working silicon there are 3 options:
1. it is intentional for whatever reason
2. the driver team is too small and unable to release driver in time
3. there is something wrong with the chip, which makes it impossible to turn some features on
Anarchist4000
Veteran
Or they're actively testing mining performance... on all cards... simultaneously.
After all, why make mining suck when you can simply devalue the currency and sell them as new "extensively tested" cards?
D
Deleted member 13524
Guest
AMD holding all the cards to themselves just to mine and profit directly from them before refurbishing and selling said cards to the users when the currency gets devalued would be the most dick and genius move from any IHV, ever.
Yeah we sold only 15k Vega cards during Q3, unfortunately. OTOH, we did have a net profit of $1B from RTG alone due to.. erm... investments.
To claim the reason can only be brought down to 3 distinct options is to seriously underestimate the sheer complexity and amount of work that goes into these products.
Windows Vs OSX? Could these results be possibly more contrived?
How would you characterize price-performance vs competition in this apples to apples comparison?
In the end I think it is true for a customer. Intentionally withholding features and performance would be new, but apart from that it is either drivers or hardware. It is too simple to explain what might be the problems at AMD, but then as a customer AMD´s problems are non of my problems.
Similar threads
