AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

The prices and performance will be more advantageous to gamers when the RX models come out, making these look too expensive, but for a professional graphics user these look like a bargain.

 

I think this has been covered but;


https://www.overclock3d.net/news/ca...ces_their_silent_loop_360_liquid_cpu_cooler/1

 

http://m.hexus.net/tech/news/graphics/107380-amd-radeon-vega-frontier-edition-available-us999/

Don't know if posted, but Vega card is $1k, water cooled $1500 for frontier edition.

Sent from my SM-G900P using Tapatalk

 

http://www.networkworld.com/article...in-for-amd-in-the-server-security-battle.html
http://www.servers-maintenance.com/2017/06/28/epyc-get-for-amd-in-the-server-security-battle/
https://www.servethehome.com/amd-epyc-supported-os-hypervisor-compatibility-matrix-launch/

https://www.techpowerup.com/234737/amd-radeon-pro-vega-frontier-edition-unboxed-benchmarked

 

Good news for those of you that want to buy a RX5xx/4xx GPU and can't find them due to miners buying them all up. If this issue isn't resolved through a driver update, then there will soon be lots of used RX5xx/4xx GPU's being dumped, and demand will disappear for new ones for mining.

If you want a GTX 1070, now would be a good time to get your's before the word gets out, the 1070 beats the 1080, and with the changes in algorithm coming it will perform even better than it is now.

Ethereum Mining Benchmark: AMD RX Series Fall Behind Over Time!

 

links;
http://techreport.com/review/32175/ryzen-pro-platform-brings-a-dash-of-epyc-to-corporate-desktops
http://www.barrons.com/articles/amd...-multi-chip-curse-says-wells-fargo-1498747117

 

Corporate IT that uses vPro can now integrate Ryzen Pro with DASH support, which is a continuation of AMD's push with their APC's previously:

AMD goes after vPro with seventh-gen PRO APUs
https://techreport.com/news/30741/amd-goes-after-vpro-with-seventh-gen-pro-apus

Except now people actually want the AMD Ryzen, so the DASH support will smooth the way to bring them in house for companies with the vPro requirement.

 

It looks like the "old" sell on good news adage is still being followed:

The Biggest Loser: Advanced Micro Devices Drops 4.7%
Tech tumbled today, and that made Advanced Micro Devices the worst-performing stock in the S&P 500.
http://www.barrons.com/articles/the-biggest-loser-advanced-micro-devices-drops-4-7-1498767327

"Advance Micro Devices tumbled today despite a positive note from Wells Fargo analyst David Wong, who wrote that "AMD's EPYC server processors may well offer better absolute performance and price/performance than Intel’s current Broadwell generation server processors, and perhaps comparable performance and price/performance to Intel’s upcoming Skylake processors which are scheduled for launch next month on July 11."

 

I saw even as the article hit that AMD was being trashed for the day. Good thing my money is all cash right now. I have been wanting to jump in but way too volatile right now.

 

What's funny is the articles trying to say AMD needs acquired to continue in business or that Goldman Sachs is saying that Epyc was already priced in. I'm calling BS. Wait for the pop at Q2 report, then a further pop after Q3 earnings (which is the real showing for Epyc test beds, TR initial sales, and Vega initial sell out, until the HBM shortage is fully abated). The 10-K in February-ish will show the early purchasers on Epyc, although H1 next year is when commitments will truly be made.

Sent from my SM-G900P using Tapatalk

 

It's looks like it's almost VEGA-Time!!...

" With hours to spare until our Vega shipment arrives from a retailer, we put together a review of the Zotac 1080 Ti Amp Extreme – it’s in editing now, and still pending completion – and tore-down the card. The tear-down is live now on YouTube, and is embedded below."

Spoiler: article and video of Zotac teardown...

Zotac GTX 1080 Ti AMP Extreme Tear-Down
http://www.gamersnexus.net/news-pc/2971-zotac-1080ti-amp-extreme-tear-down

ZOTAC 1080 Ti Amp Extreme Tear-Down

 

Vega benchmarks are too bad to believe. sad newsfor amd and gamers. Nvidia gouging is going to continue atleast for another year

 

Not if you put your money where your heart is, vote with your wallet to keep AMD in the game, send Nvidia a message to reduce their prices.

Nvidia's price "gouging" isn't because it's faster than alternatives, it because Nvidia set a price and people pay it. Stop paying that price.

A few FPS loss is a small "price" to pay to pay a lower price... yeah that's it

 

That is bull. The Vega they put out sucks for gaming but is not meant for gaming. Coincidentally, it came out a month before the gaming version. Why? To hone drivers for games like they honed drivers, software and AGESA/microcode for EPYC, which is telling them info for TR. So, what are you talking about. Hell, one article said it lost to the 1080 when it had a score higher than the 1080 in 4K firestrike ultra.

Sent from my SM-G900P using Tapatalk

 

How many times did AMD say "PLEASE DON'T DO GAME TESTING ON OUR PROFESSIONAL GPU!!!" ?

What's with the brain freeze around this issue??

It's not a gaming card, wait for RX Vega... sheesh.

 

It isn't, but they have gaming mode. They are using that to get the Rx Vega ready.

People just don't listen or think logically. Every release so far was to get things ready for the next release. Every new thing is a block.

Now, for AMD driver support, it usually took 3-6 months to get the drivers fully optimized. But they have been working on drivers since January/February this time. Then, why not use the halo project to get the game drivers honed a bit more? One month on top of internal checks, plus one month for software/game companies to optimize of they didn't get early hardware. I'm cool with that.

I'll wait for full suite testing of the frontier edition, then full suite testing of the RX Vega. I'm not judging yet.

Sent from my SM-G900P using Tapatalk

 

So, a card optimized mainly for professional use which excels in pro software gets almost no recognition, but that same card which seemingly doesn't perform as well in games using 'gaming mode' (which was made for testing purposes mainly) seems to get a bad 'rep' and can be used as a 'benchmark' for the gaming version?

How do Nvidia's pro cards perform in games?

On top of that, we're talking about what kind of drivers here exactly?
Are they ready to be used for Vega that make use of it's architecture?
I doubt it.

But, like anything else, we will have to wait and see.

Oh and, if gaming Vega cards actually DO end up A LOT better in games (and I hope they do), then wouldn't AMD incorporate this into the pro drivers for gaming mode too?

 

I read something about the top end Quadro beating the top end Titan in gaming, but not by much and the cost difference is nuts

 

That, and how much time passed between these tests and the initial drivers the cards were launched with?

Seriously, I'm seeing maybe 2 or 3 people trolling about Vega's 'bad gaming performance' already like mad and that the RX Vega won't be significantly faster at all.

That would be a rather serious assumption to make, considering that the gaming mode on Vega Frontier Edition probably isn't making use of all the hardware for gaming and is un-optimized.

I think AMD should have simply eliminated this distinction between pro and regular cards and sell the gpu's at regular prices by giving EVERYONE (consumers and professionals alike) option to switch between the driver modes.
Hardware wise, they are identical... only difference is in software.

This would mean AMD could have easily undercutted Nvidia in the pro segment for example... but it would mean the driver division would have needed time to optimize gaming performance (which is likely what they were doing with the pro side).

I guess we need to wait and see.

 

False! This has 16GB HBM2, the upcoming cards only have 8GB. There's more difference, but that is significant. Also, the two drivers are very different. It is like comparing Cuda drivers and Quadro drivers to Geforce drivers.

Sent from my SM-G900P using Tapatalk

 

Aside from VRAM amount, what kind of hardware differences are there besides the disabled cores/sections of the GPU so that you can have a lower performing version?
Gaming cards use the same approach.

But my point was that if AMD already decided to have a driver switch for the pro gpu's... why not simply remove the 'pro' line altogether and focus on having the software switch that would transition between either gaming or pro use when needed at a lower price (gaming GPU level).

 

The differences are not yet known, but I'm betting it is the primative shader utilization/bypass. A professional card may not have that at all, because it isn't the primary purpose of the card. Meanwhile, gamers don't crunch the same way professionals do. So there is a chance that isn't on the pro card, but is on the gaming version. If true, we'll see a large jump in benchmarks soon, needless to speak of software optimizations, drivers, etc.

But that is pure speculation.

Sent from my SM-G900P using Tapatalk

 

and with pics: http://www.gamersnexus.net/news-pc/2972-amd-vega-frontier-edition-tear-down-die-size-and-more

 

That doesn't look just like a copper baseplate, it almost looks like a vapor chamber.

Sent from my SM-G900P using Tapatalk

 

links;
https://www.forbes.com/forbes/welco.../www.bing.com/&referrer=https://www.bing.com/
http://pcworld.com/article/3204591/computers/amds-ryzen-pro-launch-reveals-ryzen-3-details.html
http://investorplace.com/2017/06/advanced-micro-devices-inc-amd-stock-ryzen-pro-cpus/

 

I kept noticing this jewel and got to wonder, is it done the same way as Epyc? "It offers memory encryption, secure boot and TPM 2.0 on the silicon...." Now, for Epyc, they have an arm chip if memory serves.

"AMD has had its Secure Processor in several CPUs at this point: a 32-bit ARM Cortex-A5 acting as a microcontroller that runs a secure OS/kernel with secure off-chip storage for firmware and data – this helps provide cryptographic functionality for secure key generation and key management. This starts with hardware validated boot (TPM), but includes Secure Memory Encryption and Secure Encrypted Virtualization."
http://www.anandtech.com/show/11551...7000-series-cpus-launched-and-epyc-analysis/3

If so, I wonder if that comes standard on TR as a nice touch off.

Edit: Also, it adds to reasons for wanting those chips in an enterprise environment!

 

I am not too sure of the feature sets. It does make me wonder about the present stepping. If maybe stepping 2 contains the business features as well, meaning then the PRO is a stepping 2 as well?

 

I think it's going to eventually roll out for all Ryzen "Summit Ridge" based CPU's.

AMD Readies B2 Stepping of the Ryzen "Summit Ridge" Silicon
by btarunr Monday, June 19th 2017 06:55 Discuss (41 Comments)
https://www.techpowerup.com/234476/amd-readies-b2-stepping-of-the-ryzen-summit-ridge-silicon

"AMD is readying a new stepping of its 14 nm "Summit Ridge" eight-core CPU silicon, which powers its socket AM4 Ryzen processors, according to Canard PC. The new B2 stepping reportedly addresses a lot of hardware-level errata which cannot be fixed merely by AGESA updates. According to Canard PC, the changes seem to be focused on the uncore components of "Summit Ridge." Typically, uncore refers to the integrated northbridge, which includes components such as the memory controllers, PCI-Express root complex, etc.

If the B2 stepping is mostly focused on uncore-level errata, it could mean improved PCI-Express device support, and perhaps even memory support improvements beyond even what AGESA 1.0.0.6 brings to the table. Canard PC reports that it hasn't come across any CPU core-specific errata being addressed with the B2 stepping. The glaring FMA3-related bug has been patched through BIOS updates, and most newer batches of socket AM4 motherboards come with the patch pre-installed."

 

I've been reading some comments from people who proposed why Vega FE performed as it did on games, and their explanation simply said: 'tile based rasterization was not even being used'.
On top of that, the drivers used for FE gaming mode are simply not optimized for gaming - but rather serve a testing purpose (big difference when you take into account this early release of Pro cards of course).
Of course, we already covered this bit.

I am very doubtful that there are hardware differences between RX Vega and Vega FE apart in VRAM size (unless they will offer RX vega with 16GB too).

AMD's usual practices in CPU's involve disabling cores to get lower performing hardware at cheaper prices.
For example, their GPU's that are same baseline cards and come with say 4GB VRAM seem to be 'unlockable' to 8GB with a bios flash sometime as they found it a lot easier to simply disable half the RAM and keep the rest of the card as is.

Considering that Vega is relatively expensive due to HBM2, etc. what are the chances that AMD would be specially designing Vega GPU's in gaming and Pro versions as opposed to save themselves time and money and just make them identical hardware-wise (with only differences being in VRAM sizes depending on how much VRAM one needs/wants) and switch between drivers?

We've seen Polaris based Pro GPU's that are different hardware-wise and sport SSD's on them for example (do correct me if I'm wrong though).
They didn't come out until well after Polaris first debuted and I guess you could easily say that those could really be considered 'different'.

And if it does turn out that Vega FE and RX versions are basically the same cards that operate on different drivers, then I think AMD would be better off pulling a Ryzen on Nvidia by offering ALL Vega cards at consumer price levels with ALL cards being able to switch between pro and gaming modes (fully optimized and utilized of course on both).

These distinctions between pro and gaming versions are quite frankly ridiculous.
Pro cards are usually based off gaming GPU's... they are simply priced higher due to the market they target.

 

It is a vapor chamber, yes.

 

Go read what Chew has mentioned in XS forums. B2 will not be on Ryzen. It is used for Epyc and TR so far. No word on it being used in other chips beyond those. This makes sense because of the added I/O and memory channels, with a small possibility of gearing the ram at 1:1 DR, instead of the single rate it currently uses.

 

http://segmentnext.com/2017/07/01/amd-vega-gaming-gpus-capsaicin-siggraph/
https://www.forbes.com/sites/moorin...ecured-a-spot-in-the-datacenter/#14b7ada12c91

 

Ignore the segment next article. They are also doing TR at the event, whose release is two weeks later for purchase. It clearly shows a good size presentation and will tout the Alienware system, etc.

Sent from my SM-G900P using Tapatalk

 

It's there in the announcement, it's just not labeled as RX Vega Gaming:

"On July 30, learn about our newest additions to the Radeon Pro family featuring the cutting-edge Radeon “Vega” architecture. Hear from our partners on how AMD’s product lineup – Radeon Pro, Ryzen Threadripper, EPYC and Radeon Instinct – is revolutionizing the creative pipeline, leading industries from entertainment to engineering into a new era of precision and performance."

 

I think their distinction is they want "RX" not "PRO" Vega info, I am not sure of the distinction of the two.

 

One for you guys to ponder;​

https://browser.primatelabs.com/v4/cpu/3284583

 

I am so confused, it scores lower than a Ryzen 5 1600.

 

This is what I do for you guys, confuse the hell out of everyone. Don't shoot the messenger but I will dig and dig some more for you guys.

 

Notice how much Nvidia attributes to tiling. Literally over 25% performance. Keep that in mind for them saying the RX cannot keep up with a 1080 Ti.

 

The only thing I can say to that is the repeat of what we had before.
Unverified results using software that's not properly optimized.
Even sisandra updated itself to support Ryzen and future processors, but its definitely not optimized for infinity fabric.
Geekbench is now suffering from the same thing.

Someone else posted results of alleged 'threadripper' which scored identically to 1800x.
Geekbench is easily manipulated on top of that.

 

I did that too, it seems GeekBench gets better results with higher core clock and lower amounts of threads. Since we know the Intel people will claim a win here I have to get the info out ASAP so it can be gone over.

 

Geekbench isn't providing even close to real life results here.

Are the benches crippled? Look at the scores, they are absurdly low, even by mid-tier CPU standards. I don't even know where to start...

 

Biostar Racing X370GTN Mini ITX Ryzen Motherboard Review + Linux Test

 

AMD confirms Radeon RX Vega will launch at SIGGRAPH 2017
AMD will launch Radeon RX Vega at SIGGRAPH between July 30 and August 3 in LA
By: Anthony Garreffa | Video Cards News | Posted: 2 hours, 31 mins ago
http://www.tweaktown.com/news/58253/amd-confirms-radeon-rx-vega-launch-siggraph-2017/index.html

 

Radeon Vega: Frontier Edition Review | AMD's Enigma

 

Glitch to test for... x370, other Asrock motherboards maybe, other maker motherboards maybe not

Asrock X370 Taichi Adventures: Timing Troubles (BIOS 2.40 AGESA 1006)

 

This could be the memory too. The XMP settings could be an issue. but yes, everyone should keep an eye out.

 

Here is some wishful thinking;
https://browser.primatelabs.com/v4/cpu/3269216

More realistic, and WoW, if true competes against the top 7900x;
https://browser.primatelabs.com/v4/cpu/2508840
https://browser.primatelabs.com/v4/cpu/3292707

 

LMAFO! Those are funny. I think someone pro-Intel posted the old results and now we have pro-AMD folks at it. AMD needs to hurry up.

 

Some more leaks, if you call it that (now those R15 multi-thread scores are awesome);
http://www.cpu-monkey.com/en/cpu_family-amd_epyc-36