I can't remember a since GPU gaming card consuming 375w
-
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 **
- Thread starter TNA
- Start date
- Status
I can't remember a since GPU gaming card consuming 375w
Not really, the article is highlighting the difference between the polygons in the scene (so all the models that make up the set) and the polygons that are visible. So it is correct. Also I'm certain back face culling doesn't affect total polygon count because it is one polygon.
Come on, fess up, we're not gonna judge: did you accidentally type your password in the username box while registering?
I didn't, 1Password did
Have you seen tests showing Radeon RX vega consuming that much?
These Vega GPUs are going into an iMac, if they consume over 300W, and push out that much heat Apple wouldn't even consider them.
Just like Apple ignored Hawaii and Fiji based cards.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
....
There's never been that much of a clock improvement between an air cooled card and an identical water cooled one so I'm not gonna expect that sort of difference between two identical HBM chips.
To me it looks like a blatant rip-off, They've sat watching Nvidia do it over the last year and decided to do it themselves.
Whatever they charge I'm not paying more than whatever we have to pay for whichever Pascal chip it is closest too when looking at an overall performance summary similar to what TechPowerUp do.
That's the air cooled version. 1800Mhz on Liquid version of bust!
Take the Frontier Edition specs and just have 8GB of HBM2 instead of 16GB; and potential clock speed bump since Raja said the Radeon RX would be faster.
There's never been that much of a clock improvement between an air cooled card and an identical water cooled one so I'm not gonna expect that sort of difference between two identical HBM chips.
To me it looks like a blatant rip-off, They've sat watching Nvidia do it over the last year and decided to do it themselves.
Whatever they charge I'm not paying more than whatever we have to pay for whichever Pascal chip it is closest too when looking at an overall performance summary similar to what TechPowerUp do.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
Yes they do, The Titan Z was promoted as the worlds fastest gaming card. I watched one video where Tom Peterson or whatever his name is said exactly that, "The Titan Z is the worlds fastest gaming card".If you've followed nvidia pr since the original titan's release you'd know nvidia don't specifically advertise them as gaming cards, that being said they still stick geforce gtx on top of the card in led's so doesn't really matter what they say.
Isn't 200Mhz the difference between a reference 1080Ti and a Hybrid Watercooled one from an AIB?
One can dream, since that 500-600£ increase over Air is sick joke if true.
The titan z was barely even promoted at all by nvidia, mainly because of the ridiculous £2500 price tag. It came out with no review samples and basically faded into obscurity. It had nearly a triple slot cooler and even with that the clocks on the cores had to be drastically reduced so the cooler could deal with it, total joke of a card.
The titan is a gaming card regardless of nvidia's stance on it.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
That may be but we are talking AMD here, the Fiji cards came with just 50mhz more than the air cooled card's so I can't see these being much better.
I know it was a rip off that sucked due to it not being cooled well enough, I was simply pointing out that yes Titan's are promoted as gaming cards as the Nvidia Website shows.
http://www.nvidia.co.uk/gtx-700-graphics-cards/gtx-titan-z/
They all do it
http://www.nvidia.co.uk/gtx-700-graphics-cards/gtx-titan-black/
AMDs offical slide showed 300w and now we see products listing 375w. That 375w most likely comes from AMD product blurb for retailers. Then we hear rumours from MSI I think it was about how power hungry Vega is.
The apple information is interesting and on the face of it would trust what you are saying but until we get a better idea of the Apple products I'm not sure that helps much. At least 1 of the GPUs going in the imac is a cutdown version. Do we have office clock speeds or performance numbers? Again, I can imagine Imacs get lower clocked parts to fit the power envelope. It is also possible that AMD hand picks the best core for Apple, like they did to get the Nano at much lower power than the FuryX. Possible we also see a new optimized process in time for Apple. Nvidia is using a new custom optimized 16nmFF process named 12nm from TSMC, GF might have something similar.for.AMD for later in the year. I also expect Apple is increasing the power envelope allowed. Since the new Mac pro is so far off the new imac is supposed to be filling that gap.
At least we will know more details soon enough. I'm not trying to.convince anyone that Vega clocks are 1200MHz, only putting out some caution.
Soldato
- Joined
- 27 Dec 2009
- Posts
- 2,727
- Location
- Gillingham, Kent
Do AMD and Nvidia use the same method for quoting TDP? I vaguely remember hearing a few years back that Nvidia quote "typical usage" while AMD quoted "absolute maximum" for TDP - is this right or am I remembering something that didn't happen? Just wondering if we're comparing Apples with Apples when looking at an AMD listed TDP and comparing it to the Green Team's one.
MSI never said it was power hungry. Just a quote taken from context just saying it needs a lot of power. Because a AIB tends to overclock the gpu so it could be in reference to that seen as he later said depends on performance you get back which was a reply to some one who said that doesn't sound good. So to me that is not MSI saying its power hungry at all.
LN2 overclockers could say JEEZE 1080ti needs a lot of power. Does this mean the 1080ti is power hungry?
The Original AMD slides show 225W not 300W. Even the latest Instinct MI25 slides on its release only shows "<300W"
Also the Apple specs can easily be calculated from CUs and TFLOPS.
The cut down Vega card going into the iMac is 56 CUs, and 11 TFLOPS, 8GB HBM2
That puts the cut down Vega card at around 1550Mhz also above your "cautious" 1200Mhz.
With an Option for 64 CU unit with 16GB HBM2.
https://www.apple.com/ie/imac-pro/specs/
It's the similar reduction in stream processors and compute units used between Fury and Fury X.
Considering Raja has stated on record, that the Radeon RX Vega will be faster in gaming than Frontier Edition it makes no sense what so ever that it's cut down, or clocked lower than it. FE = 13.1 TFLOPS = 1600Mhz
I don't see how you're being "cautious" even mentioning 1200Mhz for the gaming card. It makes no sense.
We know Vega tops out at 64 CU, and 4096 Stream processors, at 1200Mhz it would be drastically slower than the Frontier Edition, topping out at 9.8 T FLOPS. Which as said before is supposed to be slower than top Radeon RX Vega according to Raja; and be slower than the 56CU Vega card in the iMac.
We also know that Instinct MI25 is at 1500Mhz, and not once has an enterprise card be clocked far higher than the top end gaming/consumer equivalent.
Last edited:
No they don't use the same method, I can't remember how either side calculates their TDP though.
The Original AMD slides show 225W not 300W. Even the latest Instinct MI25 slides on its release only shows "<300W"
Also the Apple specs can easily be calculated from CUs and TFLOPS.
The cut down Vega card going into the iMac is 56 CUs, and 11 TFLOPS, 8GB HBM2
That puts the cut down Vega card at around 1550Mhz also above your "cautious" 1200Mhz.
With an Option for 64 CU unit with 16GB HBM2.
https://www.apple.com/ie/imac-pro/specs/
It's the similar reduction in stream processors and compute units used between Fury and Fury X.
Considering Raja has stated on record, that the Radeon RX Vega will be faster in gaming than Frontier Edition it makes no sense what so ever that it's cut down, or clocked lower than it. FE = 13.1 TFLOPS = 1600Mhz
I don't see how you're being "cautious" even mentioning 1200Mhz for the gaming card. It makes no sense.
We know Vega tops out at 64 CU, and 4096 Stream processors, at 1200Mhz it would be drastically slower than the Frontier Edition, topping out at 9.8 T FLOPS. Which as said before is supposed to be slower than top Radeon RX Vega according to Raja; and be slower than the 56CU Vega card in the iMac.
We also know that Instinct MI25 is at 1500Mhz, and not once has an enterprise card be clocked far higher than the top end gaming/consumer equivalent.
Yea but remember "Overclockers dream" which D.P keeps bringing up when even some one mentions what AMD or Raja has said lol. Can't trust what AMD say can we.
I dunno, D.P himself refused to believe the Frontier Edition is like a Titan card, and he kept saying it's going to cost 2-3K and that you'll be able to SLI Titan Xps for the same cost as one Frontier Edition.
Now he's going on about a 375W GPU, and that the gaming cards will likely be 1200Mhz, compared to the Instinct's 1500Mhz.
Also I don't see him complaining after Jensen mentioned Pascal would be x10 faster than Maxwell. In fact he even tried to justify that bogus claim
Even after Jensen later came out and said he used "CEO Maths" to pull out that number.
I dunno, D.P himself refused to believe the Frontier Edition is like a Titan card, and he kept saying it's going to cost 2-3K and that you'll be able to SLI Titan Xps for the same cost as one Frontier Edition.
Now he's going on about a 375W GPU, and that the gaming cards will likely be 1200Mhz, compared to the Instinct's 1500Mhz.
Also I don't see him complaining after Jensen mentioned Pascal would be x10 faster than Maxwell. In fact he even tried to justify that bogus claim
Even after Jensen later came out and said he used "CEO Maths" to pull out that number.
vega FE is $2-3K, I always stated prices in USD because there is less FOREx issues, so I got that pretty darn close.
I assumed from AMD's marketing blurb that they were aimed at professional development, GOV and deep learning. Turns out that may not be the case and the FE is just a Titan competitor. But this isn't at all clear from AND a marketing madness. They claim the gaming Vega will be faster for games, i.e the FE comes with pro driver's. They markets for full and ha!f precision teraflops and talk ability deep learning in their slides. But it seems it won't come with professional drivers is just aimed at Titan's luxury gaming market.
The 375w is being posted on numerous retail sites and AND themselves had 300w down which is bad enough! That is a hard fact.
As for Pascal being 10x faster, all the AND fanboys purposely took that out of context. The context provided in the slides was very clear and in actual benchmarking Pascal was indeed 8-12x faster, which is why companies were happy to at $10-12,000 per card. The math was very clear and presented in the slides,e.g comparing half-prrecision to full precision nets you as 2x increase off the bat. Nvidia's statements about Pascal were all accurateas far as I can tell. E.g. They said the 1079 would be faster than the 980ti, it was.
And whatever happened to Polaris being 2.5x as efficient as Fiji? Of yeah it was in some nonspecific scenario that AND didn't even explain. When Nicosia has made such statements there is typically an explanation to go with it and benchmarks to support it
Very clear what ND were referring to,their slide breaks down the numbers
http://wccftech.com/nvidia-pascal-gpu-gtc-2015/
Last edited:
Talking Rubbish mate! The FE is $1200-1800 listed right now in the USA. That's not 2000-3000; and you cannot SLI Titan Xps for the Price of one FE.
Dude more rubbish, Jensen himself stated he used CEO maths to get to x10 performance.
https://youtu.be/Tb7ZYSTYHbw?t=4m39s
lol
During GTC he then stated 2x performance per watt compared to Maxwell.
Last edited:
The faster Vega FE is $1809 without taxes, $2160 with UK VAT.
Yeah, I expected FE to cost more because I expected the FE cards not just to be some Titan competitor. I guess AND are just not that confident in that market yet.
When even explained the math, there was no trickery about it, completely transparent what he was referring to. The numbers were published clear as day. Only AND fanboys tried to make it out as something is wasn't.
Yeah, I expected FE to cost more because I expected the FE cards not just to be some Titan competitor. I guess AND are just not that confident in that market yet.
When even explained the math, there was no trickery about it, completely transparent what he was referring to. The numbers were published clear as day. Only AND fanboys tried to make it out as something is wasn't.
Dude, stop talking so much rubbish.
Now you're trying to justify the AiO card is 2-3K. It's not near it, especially not the normal FE card.
Also I linked the video on his CEO maths, he even jokes about it since it's not factual and representative at all. You're talking about fanboys but somehow state it as fact....
He's literally half chuckling in between it all.
- Status