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

mobile 3000 CPU are of zen+ and thats garbo, they got nothing over intel in the laptop space yet. if only TSMC 7nm was exclusive for AMD and AMD puts out mobile CPU in 7nm, it'll cause a huge upset. intel got real lucky that there isn't enough volume and all the 7nm must go towards server first, then desktop.

@Papusan navi still on GCN so it'll be average, nothing we haven't already seen in terms of performance. whats good is maybe the price/performance value thats about it.

i'd put a good bet it'll be on par so you're right might not cause enough damage to intel. we just have to wait for the review to really see how well it can do.

 

AMD has been clear that the Radeon VII is the top end card for 2019, and that the first Navi GPU will be under the Radeon VII performance. That's why speculation outside of those limits doesn't make sense right now.

The AMD announcement events are coming soon enough to see what AMD is really going to release. Avoiding all the rumors will make it easier to be happy with what AMD actually delivers vs what has been rumored.

 

I do not expect AMD to be delivering a broad market knock out punch out of the box. I think Epyc will be their main hitter, out delivering, at least in the beginning, well beyond Xeon. The 3000 series will put them on par if not slightly ahead of Intel current offerings. Lastly they will hold a bit on taking the consumer graphics crown. Maybe later on place a challenge to it, but not at first.

And undecidedly Threadripper, makes me so sad not having any news yet at all! Need less to say I feel like a team player that has been cut without even a thanks for playing, I was so looking to join in next season.

 

Well, there were a couple things that had to have happened. During the H1 of this year, specifically Q1, TSMC had an extra 10-20% 7nm volume available and to my knowledge, AMD did not take them up on it. Now, that very well could have been because AMD Navi needed to be re-taped out again in November, meaning that they were not using the capacity they had available because you had a 9 month period to test, etc. Why that time wasn't used to stock up on AMD Zen 2 chiplets I do not know. Only recently did it come out that they were on a B stepping revision, so likely just not ready to finalize the silicon yet. Either way, seems like a missed opportunity. But, that means they have to produce Navi and stockpile Zen 2 chiplets at the same time, with limited fab capacity, all while we are headed into Apple season on spinning out those new ARM chips, likely on 7nm+ (could be 6nm, but most likely 7nm+).

This also explains the push on TR.

At the same time, if they are doing the leaked pricing from the Sapphire rep, I'm sitting out until next year. Had the AdoredTV leak on pricing showed up, then I likely would have gotten one. But that is also because I plan on replacing my 980 Ti with 1080 Ti/2080 levels of performance in the $400-450 range. That meant either a new AMD card or a used 1080 Ti, most likely. But that is me.

On TR delay, see my response to Ole.

I agree, they are going to disrupt servers. Last year, Dell only planned on the three offerings. Now, they have 9 offerings and are singing AMD's praises. HPE is also going to be more of a partner because of buying Cray, which has Shasta and Frontier supercomputers powered by AMD, which should allow AMD to take the upper slots on the Supercomputer 100 list. It's a good thing Intel felt pressure to put out the AP chips, because they REQUIRE liquid cooling. Because of that, Cray developing and using custom water blocks for the AMD supercomputers doesn't seem outlandish, which really helped.

As to graphics, we have to remember the timeline. They pulled the GPU budget and gave it to the CPU team to create Zen. Zen released a couple years ago (2017). Since then, they took the profits and besides paying down their debt to no longer face bankruptcy, they heavily invested in R&D.

Now, if we assume a normal 4-5 years for development, that means the fruits of Zen financed R&D (meaning the profits of Zen being rolled into R&D) will not appear until 2021/22. So just a couple years out.

Also, Vega and Navi were the last designs on GCN. Beyond that, they are switching to super-SIMD and beyond (including chiplets, which all companies are doing on GPU development). Considering the engineers just want to be done with Navi and move on, it's a good bet that Navi will rely on pricing to move it and get market share, which has been my argument for awhile now (everything sales if the price is right).

Interestingly enough, all the press said the leaked prices were too low. Then, when the higher rumored prices come out, the interest drops off a cliff. Why? Because the media helped create a self-fulfilling prophecy. They said do the higher margins (which there very well may be other reasons the prices were higher, like the bill of components), then promptly turn to say it is not competitive without those prices. Which is it?

Either way, I'm not banking on AMD graphics getting back in the game until 2021/22.

 

I mean they (AMD) was't very interested in computer gaming GPU's and instead put most of the $$$ on consol gaming. And TR was a luck due their engineers had fun.

 

I again have little issue if TR is on a push, but get it back on the map then. Even if it is in 2020....

 

Very true on TR. But I have a different argument on console gaming. That was part of their embedded products. During the period before Ryzen, back with the Bulldozer and sequential processors that literally just got owned in about every way imaginable, their embedded solutions, like the jaguar modified cores for PS4 I believe it was, etc., kept them afloat and out of bankruptcy.

Because of that, embedded console partnerships kept them alive when death was knocking on their door. The partnership on creating Vega with Apple and Navi with Sony, and the next gen with Microsoft helped to spread the R&D costs, which kept them alive for so long.

Now, with Ryzen out and making them profitable again (even if barely), you are seeing more money put into R&D, which the results of that will likely come in the generation past the Microsoft partnership. Will they have partnered with a console or embedded partner again? Possibly (because why not?). But one thing is for sure, they will finally get the budget for making cards back up, possibly to levels like back with the 6970 cards, which would be great.

Even with that, AMD has a mind share problem, and I don't believe Navi, even if priced right, will fix that. Because of that and datacenter being the faster growing while giving better margins (AMD is a powerhouse for those uses, even if not as good at gaming), AMD is focusing on business applications. All we can hope is that gaming will become competitive again, if not by AMD, then by Intel. But even Intel is trying to get the lucrative datacenter market share, so who knows...

 

if TR3 is gonna get pushed back to 2020 they may as well have it come with zen3's changes and 7nm+. SMT3/4 and better node is welcomed.

 

I don't know if that is viable. TR3 at possible 32 cores on 2 active CCX's seems where they may need to cut it for now. Meaning the last of the TR series for now. I mean really for the near future what more do you need? Especially since the 4000 series and further will offer more advances.

Edit; I mean really, a home 32 core TR3 at 5 GHz may be well over 8,000 on CBr 15

 

5ghz seems far fetched, i'd opt for 4.6ghz on all 32 cores already seem too high of an expectation, with chiller definitely possible though.

@ajc9988 the new geekbench benchmark says it all, with that ipc improvement looks like it has definitely caught up to intel in ST and we are talking about pre mitigation patches. let see if we can find some 16 cores ryzen laptop, i'd be interested in one.

 

That is impressive on the score.

Also, you have to remember, the CPU dies are shrunk more than half of the original (72-76mm^2 vs 212mm^2), although you have the central I/O. But since the I/O is spread along the center and the core dies are spread to the corners, likely, for the 4x8-core dies (instead of two in each catty-corner), the heat should be well dispersed allowing for at least your 4.6GHz. Also, I agree, with a chiller, it would definitely be 5GHz. But, that also means the boost should allow for at least 5GHz on 4-8 cores at a time (up to one or two cores boosting per core die, potentially). Also, they revamped precision boost, theoretically reducing the amount that it has to go down from its peak boost, likely through better core and thread management and better boost algorithms.

For Zen 2, the consensus, generally, is that boost can do as well as a manual overclock, but better since it also allows lighter threaded workloads to use the higher single or lightly multithreaded speeds. Overall, that means better performance and less tuning (better for noobs, less fun for enthusiasts). But I didn't consider the simplistic clocking of Intel as fun as first gen Ryzen OCing (because you had to work for it for a change rather than just getting a binned chip and going). Now, Zen + seems to be less fun do to a better boost algorithm. That means Zen 2 might be even less fun, but since it may take some crowns, benchers will make the move.

I just wish M$ would stop messing up their OS more. I have some decent ranting in the Windows 10 page worth a read on 1903 ( http://forum.notebookreview.com/threads/windows-10.762434/page-565#post-10915156).

 

os sucks, we can't take advantage of more cores and optane SSD because of software. all enterprise software can make use of those just fine and we are still stuck on crappy windows.

one thing i'd need to correct you is that, it is not less fun for enthusiasts. imagine having to go through 64 cores tweaking voltage for each core? this is the way going forward and as long as AMD is honest and transparent unlike intel, we are in good hands.

 

I just don't know if they will allow for a per core voltage tweak. Sure, a per core frequency tweak. But if I cannot then change voltage per core, I just don't know.

And, depending on their algo, there may be that or may not. That is my fear. If it is just put it in and step up voltage and let it boost, there is so little nuance to overclocking for that all while it truly will come down to binning. And that brings me to another point. We don't know if AMD will have a wide spread on their chips.

Edit: To be clear, I'd still get the chip and pay the Silicon Lottery premium for the top bin.

In fact, I likely may save my 1950X because no one else is able to really get [email protected], so if I ever do LN2, I really would want to go full pot on this if I can (IMC cold issues permitting).

 

Unless MS gets their act together, provides better performance on multi-core CPU's, at least matching the level of the best Linux performance, people will be forced to move to Linux at least as a host environment for a Windows VM's to make the most use of the bounty of high count multi-core CPU's, even for personal use.

MS's failure to provide stable upgrades for years over years is really finally getting through to people's last nerve, even those that have been stoically supporting MS throughout the Windows 10 debacle.

I don't see MS figuring this out and successfully doing something about both these problems before it's too late.

 

After a clean install of 1903 and using the newest versions of CPU-Z and HWInfo, this is what I get:



Now, there is a chance that CPU-Z needs to work on reading the bus speed because I have previously complained and they were able to fix the variance to a degree on their end. And considering older versions are slightly closer to the set frequency, there could be issues in both of their most recent versions. But with the RTC being off, etc., the problem could lie in part with changes to the OS as well. I can identify the problems, but cannot deep dive into them (my own technical limitations).

 

Nvidia's lodged trademark applications for 3080, 4080, 5080 in the EU which seems to be to deny the Navi parts those numbernames.

https://www.tomshardware.com/news/nvidia-amd-navi-3080-4080-5080-trademarks,39435.html


Why this was not done months ago (along with red herring applications to keep the media) with at least preliminary applications years ago when Nvidia restarted with the 10x0 naming, or when AMD blindsided Intel with "X399", or when the first suggestion of a Navi 30x0 naming scheme was rumoured... I would have thought a $100bn market cap megacorp would be a bit more proactive in defence of their marketing IP.

Or AMD could just go back to the old days and call it "2080+"...

 

" If the leaks are right, the chipmaker might once again switch it up with Navi."

AMD never said anything about using that numbering, it's all rumor / speculation, and if AMD didn't reserve those number names then that would suggest that AMD never had the intention of using them, or even wasting time officially creating a "red herring".

Again the rumors are just a bunch of BS to pass the time until the official word comes at release.

 

Also, many countries allow for unpublished applications. Overall, names don't matter, what comes out and the performance thereof does.

 

It is kind of hard to just TM a number, now GTX-3080 or RTX-3080 etc. is not much of an issue. Even if AMD's offerings were faster than nVidia's I doubt they want the naming convection for direct comparison, at least not yet.

 

A little under 4 hours until AMD's presentation tonight at Computex, very exciting stuff! I'll be tuning in to see what they are bringing to the table.

 

Update: The AMD video link now goes to an archived version, and AMD left a bunch of idle time in the recording, so you'll want to jump to 29:00 to start, perhaps all the way to 41:40 when the AMD presentation begins. Also, something is odd in the playback now, IDK if it's just me, but if you notice halting playback it's not just you.

Update: Twitch archived the video @ https://www.twitch.tv/videos/430443641

AMD at Computex 2019
AMD
Started streaming 6 minutes ago
Watch AMD President and CEO Dr. Lisa Su's keynote at Computex 2019 in Taipei, speaking to the next generation of high-performance computing.


AMD Twitch doesn't have a feed - yet? - but AMD Twitch Live Chat is active...
https://www.twitch.tv/amd

"NeedMoreSleep: I'm just gonna watch on YT and chat on Twitch because you guys are always hype af"

YT has Chat disabled...

Update: The AMD twitch feed now has Live video feed too

-------------------------------------------------------------

Here's a couple of Live feeds to try, it's Taipei time Monday 27th 10am, which is US 7pm Pacific time, 10pm Eastern Time.

Here's a new channel, not subscribed before, so make sure you have an alternative:

AMD Computex 2019 Keynote Livestream + Live Reaction & Discussion
Jays Tech Vault
Scheduled for May 26, 2019
Today we react to the official AMD Computex 2019 keynote from AMD's CEO Lisa Su live on stream. Come join as we take a look at what is released and what is not.


AMD Computex 2019 Keynote: watch the livestream...
by Hilbert Hagedoorn on: 05/26/2019 05:49 PM 13 comment(s)
AMD will open with the Computex's 2019
https://www.guru3d.com/news-story/a...-livestream-with-ryzen-3000-announcement.html

"Here on Guru3D you can watch the Computex keynote live on site. We will live stream the event from this very news item. The Computex Keynote begins on Monday, May 27, at 10:00 am local time.

• For Western Europeans in the night from Sunday to Monday at 04:00 clock if you want to see the stream live.
• For the US that PT 7 PM / ET 10 PM
• And in Asia, 10 AM Taipei Time (GMT + 8 hours)."

AMD to Livestream Dr. Lisa Su Computex 2019 Keynote
http://ir.amd.com/news-releases/news-release-details/amd-livestream-dr-lisa-su-computex-2019-keynote

"A live stream of the event will be available at: https://www.amd.com/en/events/computex. An archived version of the webcast will be available approximately one hour after the event and can be found on the AMD YouTube channel."

That leads to a webcast sign up, which I signed up for, but I haven't received a link / email yet. Maybe that URL turns into the Live feed later...

Some others might pop up as time gets closer:
https://www.youtube.com/results?search_query=AMD+Computex+2019+Keynote&sp=CAI%3D

 

Ryzen 9 3900X - 12C/24T - $499 Ryzen 9 3800X - 8C/16T - $399 Ryzen 7 3700X - 8C/16T - $329

Read more: https://www.tweaktown.com/news/66054/amd-ryzen-9-3900x-12c-24t-chip-cost-499/index.html

No 5Ghz chips out of box..

 

And no 16 core.

As the saying goes... If you're being chased by a lion, you only have to outrun one other guy, you don't have to outrun everyone.

Why beat Intel any more than they have to? The 12 core does just that... for the $499 price all the leaks said the (already not top tier) 16 core "3800X" would be

#sAMDbagging

 

Welp looks like the prices were right. The $499 12 core is the only excting thing I see here and I would be interested only if it beats an overclock 9900K at gaming. Other than that meh. 7nm hasn't brought huge clock gains, still no 5Ghz, I hoped for more. If that 3900X turns out to be a beast though for gaming, I would give it a chance for sure!

 

Upset, no 16 core and still no 5 Ghz. Also TR3 is still out in the air. This leaves no immediate upgrade path for TR owners at all.

 

Yeah, the boost clocks seem disappointing, but the 15% IPC is no joke! I'll work out the math in the morning, but that is around 8%+ IPC over Intel.

There is so much more I want to know, like seeing XFR and PBO performance.

But seeing the gaming deficit being closed at stock is nice.



Sent from my SM-G900P using Tapatalk

 

I'm not at all that disappointed TBH, the clockspeeds have been brought up, the the core count increased on the consumer line at a decent TDP. If history repeats itself with AMD, they clocked these chips to near max, with 100-300mhz on the table for OC on conventional cooling. That means around 4.9Ghz boost? That isn't bad at all. The 12 core could be a killer chip.

 

Quick notes on the keynote speech today:

I'm happy that AMD is still making headway, but at best, they're still catching up to Intel circa late 2018, (and only if independent testing actually reinforces what they presented today in about a month and a half).

The datacenter stats presented were interesting and may prove to be their savior for the next year, depending on what Intel will be releasing in the next few quarters too, of course.

The lower prices, the 'matches' Intel's current offerings and the lower TDP are not the building blocks to 'better', let alone superior.

Either way, I like how the ball is hit from one court back to the other with each iteration from both companies.

Dr. Lisa was slightly annoying to listen to. Out of breath and sounding like she didn't quite believe in what she was saying either (or was that just me).

For 7nm products at limited power, the performance jump indicated is in the expected range. The better choice would have been screw the TDP and let those cores actually fly.

'Supreme Performance' at a price isn't a thing. Except in the minds of marketing ' professionals'.

 

I'm definitely interested to see the 12 core on water! I'm wondering how far it will go considering it's showing against the 9920X.

Now, the caveats. The Navi was shown inn strange brigade, likely using Vulkan which heavily favors AMD.

The CB20 score was impressive against the 9700K at nearly 33%. But AMD SMT is exceptional at that type of workload. But they also said equal single core performance at stock. Granted, LUUMI just uploaded him overclocking a 9900K on water to 5.8GHz on the R0 stepping.

For PUBG, it showed equal performance in the title (maybe slight edge to AMD), but with the KS coming out, it may outperform, granted at a huge premium.

For the 3900X, they used blender, a known heavily MT program, but which stores a lot in cache (so avoids a deep exam of memory performance) and which even the 2990WX saw little deficit due to multi-chiplet design.

It's important to note those things to make sure people see the full picture. But the 12-core at 3.8 base/4.6 boost does look like a great offering, priced the same as the 9900K and possibly less than the 9900KS.


Sent from my SM-G900P using Tapatalk

 

I hadn't seen that video by him yet. I saw his video of the 9900KF calling it a TURD essentially. I will check out that 9900K R0 stepping video now! 5.8Ghz on water, WTF! I am exctied to see 3rd party reviews of the 12 core Ryzen chip. I hope it can overclock and maybe AMD didn't steal all of the fun this round. The 12 core has my interest.

 

He put it up around the time of the AMD keynote. I'm a subscriber but haven't watched that one yet, but felt it should be mentioned and shared.

I'm also excited for the reviews! Solid showing. We'll get more in 2 weeks at E3. Then another 3 weeks for reviews.

I'm also wanting to know more about RDNA, because rumors had it as a GCN core. Has me very curious!

Sent from my SM-G900P using Tapatalk

 

IPC IS CAUGHT UP BOYS. TIME TO REJOYCE

 

Possibly, but only to Q4 2018 levels.

See:
http://forum.notebookreview.com/thr...ga-polaris-gpus.799348/page-593#post-10915573


See:
https://www.tomshardware.com/news/amd-third-gen-ryzen-7nm-launch-intel-cpu,39449.html

 

So, 5.8 was just a CPU-Z validation. 5.4GHz running CB15, and that was 1.34V. He couldn't get 5.5 due to ambient temps.

In the morning, I'll work out the details of it, but it would likely take the AMD chip at 5GHz all core to match the Intel chip at 5.4GHz, which would be 500MHz over the stated boost. I don't know that it will hit that without chilled water on AMD.

But if it can do 4.7 all core, that would be around Intel at 5-5.1.

Sent from my SM-G900P using Tapatalk

 

See:
https://www.tomshardware.com/news/amd-threadripper-roadmap-removal,39460.html

No actual news, but possibly something in 10 days or so.

 

With another presentation on June 10th, AMD needed to save some announcement surprises.

AMD R9 3900X, 3800X, 3700X Specs & Price: 16-Core Held Back for Now (& RX 5700 GPU)
Gamers Nexus
Published on May 26, 2019
AMD only detailed some of its Ryzen 3000 product line specs, leaving much for E3 and later in the year. We'll go over what we have, like 3900X, 3800X, & 3700X specs and price. AMD announced the 12C/24T R9 3900X at $500, the 8C/16T 3800X at $400, and the R7 3700X 8C/16T CPU at $330.

Note also that the 16-core CPU appears to have been held back until later this year, but we do know it exists.

 

I also appreciated AMD including Threadripper in the intro roll up to bringing on Lisa Su, Threadripper is not forgotten.

Loved the title of that article "Lisa Su: You Will Definitely See More Threadrippers From Us".

AMD also didn't show the RDNA RX 5800, 5900, or half steps in between - if they exist.

We also didn't see the 16 core Ryzen 3950(?), and I wouldn't be surprised if AMD released the 8c/16t and 12c/24t first into the market, which should match or exceed Intel performance for less $. Then after Intel announces their updates, AMD can follow up with a 16c/32t knock out punch.

I hope the 16c/32t fits into most previous motherboards, but if not that might be a good reason to hold off until the more compatible 8c/16t and 12c/24t CPU's show off the x570 and PCIE 4.0 with the new Navi GPU's. Hopefully that will reduce the discomfort of shelling out $ for an x570 to get the maximum performance out of your new Ryzen 3 / Navi builds.

There was talk awhile back of some previous motherboards supporting PCIE 4.0 through software updates, but how that would work given the old motherboard chipsets?

All in all a great presentation by AMD, but it would have been doubly exciting if all the AMD new product announcements from both presentations (June 10th too) could have come out in a single presentation.

 

10nm+ is still under 14nm++ so no problem. since that late 2018 IPC won't have anything on desktop until 2021 by that time we'll have AMD on TSMC's 7nm+ which will boost IPC and frequency by a bit more. intel will likely be on 10nm+ at that time so it'll be VERY interesting to see what they bring to table over core arch with IPC improvement along with similar clocks with AMD.

its good to see Navi is on new arch, we all thought its still gonna be on GCN and that missing 5800 and 5900 seems promising. sad part is that we all know once nvidia moves onto TSMC's 7nm they'll be able to put almost 2x the compute units into it so we will be back where we are now. hopefully people will actually buy AMD's product instead of waiting for nvidia.

also about 16 cores. @hmscott @ajc9988 @TANWare there won't be any 16c yet the same reason as there isn't TR3 atm because all 8 cores chiplet gone to eypc. maybe end of the yr or early next yr.

 

@ajc9988 any in case chiller? (inside the case, built in chiller?)

 

Waiting longer to release the 16 core chips is also a good idea since they can bin them for the dies that will do the highest frequency.

 

While the higher speed, lower power, greater core counts Ryzen 3 + new architecture Navi GPU's are awesome, the x570 upgrades of memory performance and more importantly PCIE 4.0 might prove to add so much more to the performance as a package overall.

Going from HDD 100MB/sec to SSD 600MB/sec was a 6x speed up and was a noticeable difference in performance in everyday use - everyone could see and benefit from the speed up, while the follow on PCIE 3.0 1000MB/sec upgrade wasn't even 2x faster than SATA 3 SSD's.

The follow on PCIE 3.0 performance upgrades to 2000MB/sec were 3x faster than SATA 3, still not enough of a difference for most users to justify spending 2x as much for the same amount of storage size. PCIE 3.0 performance wasn't slicing down the "wall time" enough to notice in general usage, except in niche special needs where it made a useful difference, but for the average user not so much.

Now with PCIE 4.0 the first SSD is showing just shy of 5000MB/sec - pending reviews confirming - which gives almost 10x going from 500MB-600MB/sec SATA 3 to 5000MB/sec PCIE 4.0.

That PCIE 4.0 10x speed up should finally be a noticeable speed up from SATA 3 SSD's, and 2x-5x faster than PCIE 3.0 SSD's. I hope the pricing won't be outrageous...for long.

Corsair Announces AMD X570 and PCIe 4.0 Force Series MP600 M.2 SSD at 4950MB/sec
by Hilbert Hagedoorn on: 05/27/2019 09:55 AM
https://www.guru3d.com/news-story/c...-force-series-mp600-m-2-ssd-at-4950mbsec.html

Spoiler: See the article at the link - some details here:

"Corsair today unveiled the new CORSAIR Force Series MP600 NVMe PCIe Gen4 x4 M.2 SSD, one of the world’s first PCIe 4.0 M.2 SSDs compatible with the new AMD X570 chipset, also revealed today.

Unleashed by the PCIe 4.0 ready AMD X570 chipset, the MP600 reaches new heights of enthusiast single-drive storage performance, delivering up to 4950MB/sec sequential read – ten times the performance of many SATA SSDs, and fifty times faster than some hard disk drives.

The MP600’s phenomenal performance stems from the hugely increased bandwidth of PCIe 4.0 (PCI-Express Generation 4), a feature that will be made available to customers for the first time as part of the AMD X570 chipset and 3rd Generation AMD Ryzen™ Desktop Processors.

Easily fitting into a PCIe 4.0 x4 M.2 slot, the MP600’s NVMe interface and high-density 3D TLC NAND combine with a Phison PS5016-E16 controller to enable a new level of single-drive SSD performance.

Boasting up to 4,950MB/s sequential read and 4,250MB/s sequential write speeds, the MP600 is ready to become the storage center of your new 3rd Gen AMD Ryzen Desktop Processor and AMD X570 chipset-based system.
...
Availability, Warranty and Pricing
The CORSAIR Force Series MP600 NVMe PCIe Gen4 x4 M.2 SSD will be available alongside the full launch of AMD X570 chipset motherboards and new 3rd Generation AMD Ryzen™ Desktop Processors in July 2019. The MP600 will be available from the CORSAIR worldwide network of authorized retailers and distributors. The Force Series MP600 is backed by a five-year warranty and the CORSAIR worldwide customer service and technical support network."

Look at the heat sink - wow, that's gonna benefit from some active cooling blowing away the heat, and IDK how that's gonna fit into a laptop - maybe a laptop front to back dedicated channel to blow air over the PCIE 4.0 storage devices? Wild.


This performance improvement might be enough to encourage Ryzen owners to sell their old rig (and storage) and build a new PC with the new x570 + PCIE 4.0 + Ryzen 3 features + Navi with PCIE 4.0.

This performance advantage over previous generation motherboard chipsets matched with Ryzen 3 / Navi might be why vendors are making so many new (x570) motherboards this generation.

Here's another one...

Gigabyte's Aorus PCIe 4.0 SSD, Under the Hood at 5GB/s
https://www.tomshardware.com/news/gigabyte-pcie-4.0-ssd-5gb,39433.html

 

Active cooling might be recommend otherwise jokebook won't handle the heat.
I think AMD will gobble up few shares from Prosumers who would love PCIe 4.0 based SSD for faster export times.
AMD is futureproofed with PCIe 4.0 and newer tech. TDP are good too.

 

Well, I wouldn't go so far as to say "future proof", unless you are talking about 2-3 years... of course it won't slow down after 2-3 years, but PCIE 5.0 is coming down the road - Intel says 2021. AMD may wait and upgrade at some point around the same time, if there is a real advantage to be delivered.

It may be that PCIE 4.0 can provide enough headroom to not hold back performance well into the PCIE 5.0 release cycle. The PCIE 5.0 added performance headroom may not be needed immediately when it's released, as PCIE 3.0 was when it came out - it was a couple of years before devices began to take advantage of the added bandwidth.

Then there is DDR5 arriving for AMD at some point, maybe for the next generation AMD CPU upgrade there will be support for DDR5 added to the next generation chipset.

For now though, the PCIE 4.0 storage performance upgrade is a nice performance bump.

 

I keep hardwares for longer say 5-7 years. I'm seeing crappy performance on Intel CPUs after CC uCodes for skylake. IO, processing time etc.. have taken a noticeable hit.

 

Here's another great example how fine tuning the Radeon VII can get you well past the 2080/1080ti performance @ 4k and within 10% or even faster than the 2080ti in new games from the ground up for DX12:


RADEON VII Special Edition Review: The Threadripper of Graphics Cards!
Moore's Law Is Dead
Premiered May 25, 2019
This is the Radeon VII review that actually covers real questions: Does it mine well? Can it overclock to a 2080 Ti? Who is the VII for? Is it worth getting over Navi?

Spoiler: TIMESTAMPS BELOW!

Background: 0:23
Why No RTX 2080?: 4:17
Overclocking Guide: 7:00
My Test System: 11:20
Mining Performance: 12:05
1080p: 12:38
1440p and 4K Benchmarks: 13:39
Performance Summaries: 18:54
Who Should buy the Radeon VII?: 22:11

LINKS!:
https://techgage.com/article/amd-rade...
https://www.guru3d.com/articles_pages...
https://www.techpowerup.com/reviews/H...
https://www.techpowerup.com/reviews/H...
https://www.tweaktown.com/articles/89...
https://wccftech.com/the-division-2-p...

 

one issue with the story is AMD is claiming the lack of TR story took life of its own, actually no. THEY deleted it from the roadmap so they gave life to the story by not reincluding it or an announcement to the contrary. Normal corporate spin machine at work.

 

Well, there is only so much "stage time" or attention these product releases can sustain, and diluting it by also discussing new Threadripper CPU's that might be out there 6 months or more - what is the "normal" release cycle gap between Ryzen / Threadripper? - wouldn't make sense either.

If you want to be inconsolable, that's up to you...

Maybe this will cheer you up...

AMD is the New Intel, and they will be a better Market Leader!
Moore's Law Is Dead
Premiered 5 hours ago
"I am getting ahead of everyone here and pointing out how the 3900X is 10x better than anything Intel would have given us by now. AMD is the new CPU market Leader, and I am worried about Intel Long Term..."

Imagine how much better Threadripper 3 is going to be using what we've seen with Ryzen 3, plus there may be even more improvements...

Good things come to those who wait.

 

Not asking for info other than it is still on the roadmap, where is that divulging info that was not previously officially presented?

 

AMD CEO confirms Threadripper is alive, well, and 'moving up'
It sounds like Computex will be phase one of AMD's announcements.
https://www.pcworld.com/article/339...threadripper-is-alive-well-and-moving-up.html

"Though AMD CEO and President Lisa Su didn’t mention the company’s Threadripper processor in her Computex presentation Monday morning, she told reporters that the company hasn’t discontinued it.

Su wowed a keynote audience here on Monday with the disclosure of the company’s first 12-core Ryzen 9 processor, the first branding of the upcoming “Navi” processor, and more. But any omission of a key product raises questions of its viability, and Su put those questions to rest.

“You know. it’s very interesting, some of the things that circulate on the Internet—I don’t think we ever said that Threadripper was not going to continue—it somehow took on a life of its own on the Internet,” Su said, speaking to a small group of reporters following her keynote.

“You will see more [Threadripper] from us; you will definitely see more.”"

That sounds to me like a definitive statement as to Threadripper continuing forward as part of AMD's roadmap for the future.

https://www.google.com/search?q=Lisa+Su:+You+Will+Definitely+See+More+Threadrippers+From+U

 

Of course, that video guy has an opinion. Yet nothing has actually changed right now.

I have no doubt that Intel will not stay at the top, indefinitely.

But they are far from rolling over and playing dead today, or in the near to medium future that we can all see.

AMD is still pushing more cores as if they're a gift from them. They're not. They can only offer what they have though. After three years of discussing this on these forums, they have finally matched the single core performance (potentially, in some of their very near future products) that Intel offers and has offered, all along.

When AMD or anyone else offers a real advantage over proven platforms (and the security issues that have plagued Intel? Mostly, not 'real' for me in my usage, as discussed in another thread...), then that video's content will make sense.

Right now? When a 4C/8T two or three-year-old platform gives an equivalent performance to what will be released in July 2019?

Yeah; they are right where almost everyone says Intel is at; offering a new platform with no real gain in real-world day to day use.

A few more cores for half the cost of a processor will not save AMD, they'd better have something more than that, and soon.

To give a direct example, myself, why would I upgrade a few dozen systems today at effectively full cost (when all aspects of a new platform are considered, to get to where I already am?

I have no doubt that there are some out there that will upgrade just because they can. And yeah, I too will be buying the newest AMD platform and testing it for my uses too, of course. But I have a standard against which they have to live up to.

Anyone with no system at all today should consider AMD.

Buying an AMD system because of the extra cores though for the masses? Today, that is still like buying gasoline to diesel conversion kit and then watching the price of diesel surpass regular gasoline before you've paid off for the conversion.

TR3 will be interesting to see what that brings. 10 days seems like too long to wait right now (and even worse if it will not be available before this year is out...).