Ryzen vs i7 (Mainstream); Threadripper vs i9 (HEDT); X299 vs X399/TRX40; Xeon vs Epyc

Yeah, I was hoping to redirect the "Forget Insecure Intel, Let's all buy AMD CPU's!" OT posts out of that thread and bring them over here where it's on topic...

I might as well point out the NetCAT news:

CPU Vulnerabilities, Meltdown and Spectre, Kernel Page Table Isolation Patches, and more
http://forum.notebookreview.com/thr...atches-and-more.812424/page-128#post-10950250

 

what is the memory penalty of having quad channel vs duel channel on your current TR system do u know?

also is it possible for you to go 3 channels instead

 

The penalty can be very severe in some cases. Try changing the interleaving on a threadripper to use half the bandwidth on memory to do AES. It cuts performance drastically. I even told you encryption was part of my interest and showed my benchmarks on it for sisoft. So go on. You aren't worth dealing with. I've tried multiple times to explain things to you.

And 3 channels was used LONG ago. Look up old 6-dimm motherboards. That is why Intel currently uses 6-channels on their big socket server chips.

 

i forgot the numa stuff with first and 2nd gen TR due to it's design. im more looking for the memory penalty for dual channel vs quad channel, like on intel's HEDT system and in this case TR3000.

TR3000 will have similar chiplet design as 3900x so that means with higher memory latency by default compare to zen+. going quad channel means more latency than dual channel but i wish to find out how much of a penalty there is with TR3000 quad vs dual and intel HEDT quad vs dual.

 

So you will ignore the performance for a figure that is divorced from performance overall (latency), while not understanding the impact of the huge L3 cache nor the fact bandwidth impact is task dependent and keeping the cache fed. If you want to know, test it yourself. As I've said, I have tried to explain the importance of the different subsystems and their interplay to you before. The fact you are now not understanding what I've talked to you about before, while trying to ask a question so ambiguous that I cannot give an answer because you don't have all the necessary information shows you seem to have ulterior motives here.

 

lets not put word in my mouth. regardless of the penalty of architecture and who gets better advantage, i simply wish to know what the penalty is going from channel 2 to channel 4 on both intel and amd system. this is why me and @tilleroftheearth have such hard time talking to you because you keep doing the strawman argument.

i asked you because i know you have TR1 system and im asking your assistance to check memory latency on your quad channel, then test it in dual channel to see the difference. if you donno or dont wish to answer, just say so and i'll find out myself or i'll ask others. stop accusing of others what they havent do or what they may do, stop putting words in other's mouth, stop overthinking it and sometimes take what we ask and say at face value, it'll help you a lot.

 

Go look up my AES scores in geekbench 5. There is NO penalty going to quad channel, tons of penalty going to dual channel. The specific AES in that test is implemented in bitlocker, among other programs. You'll see a 12.5GB/s and around 25GB/s scores. In fact, their implementation this time lowered my mem bandwidth for that test from GB3 being 33GBps, GB4 was 30GBps, and GB5 is 25GBps while 7980XE is getting 33GBps without having as fast of ram on it's system. But that can also be effected by software optimizations as well. As i said, whether bandwidth is utilized depends on the task, but how well it utilizes memory calls to keep cache fed is also driver and software optimization related.

Now if you are talking non-linear scaling, referring to rank stress on the memory controller, etc., then that is a deeper discussion.

But since I already told you you didn't give enough information for me to answer and wanting me to guess what you are looking for for an answer, go find it yourself!

 

i always thought going from dual channel to quad would increase latency cause of overhead. kinda like SSD raid, dual core has higher single core latency than a single core cpu, bonded internet vs single loop etc etc.

are you sure theres 0 penalty or the penalty is too small to factor much

 

I looked it over, and I seem to remember this is the guy who did the original testing, but a couple of things he said that make no sense to me...

He went on about thread limiting before, but now says it's completely unnecessary. I don't understand that bit, and I would really have liked to see a comparison of how his 7980XE did against the 3900X directly, and if the 3900X (as he makes it sound) overcomes problems his 7980XE has (which also confuses me considering quad channel benefit on the latter over dual channel only on the former). He also said that "it can stream anything out the box without issue or tweaking", but then also says "if you do 1080/60 streaming it's going to run the CPU at full speed and you are going to need better cooling". That one's also pretty weird, since... first of all, duh, and second of all, if it really is maxing out the load the games should be suffering a fair amount, and I KNOW that AMD has bigger trouble with load balancing than intel does when games want high single thread usage and a large CPU load is being used due to much lower single core speeds, which... seems to be the reverse case for him? I really want to see a head to head between the two. Especially with utilization data in realtime as he tests (like an afterburner layout properly done and showing per-thread utilization).

I still do feel NVENC is the better way to go overall, though. I don't think my stance on that one is going to change anytime soon, but I'd really like to see a head to head "this one wins where this one loses" even if he doesn't understand the reason. He knows a fair bit about streaming, I will not discount that. One of the most knowledge-able people about the technicalities I've come across, but he makes some weird statements/choices to me.

 

This first one is with the interleaving setup wrong resulting in a reduction in performance (it acting like dual channel UMA on the test). This was right after I had flashed the newest BIOS and was setting things back up and had setup the channel interleaving wrong in the BIOS. This is not setting it purely up for dual channel in the way going to "game mode" does within Ryzen Master, but it is pretty close. Here is also after the interleaving was correctly set in BIOS so that it was fully utilizing the quad channel memory.




Geekbench removed the memory testing from this version of their benchmark. But you can see it vary per test on which tests are more memory sensitive and which ones are not. Here, AES-XTS, Machine Learning, and speech recognition are all HEAVILY hit by the memory settings and bandwidth. Lesser effected tasks, as coded by Geekbench, include text rendering and navigation. But beyond that, such as text compression, horizon detection, image inpainting, and HDR, you see it trail off, being within 1/6th of the score. Those likely are more about the boost the bandwidth gives by keeping the cache fed, but not so much that it is starved like the other two categories I mentioned. Many of the rest are margin of error type stuff, where so long as the cache is efficiently fed, it is negligible (except maybe camera, but I haven't dove so deep in analyzing this new benchmark yet to rundown all of its nuance).

Single core performance is different. Why? Because it doesn't have to split the bandwidth between the cores. It only has to keep the active core alive and well fed, not splitting up the bandwidth between many cores (hence why I mentioned the bandwidth per core previously before you attacked me and I wrote you off).


https://browser.geekbench.com/v5/cpu/compare/25528?baseline=25528
As can be seen is there are some effects from quad channel relating to the overhead you mentioned, but the vast majority are de minimis. Now, this also depends on how you have the ram setup for interleaving. That is where a discussion of channel and rank interleaving comes in. Let's say you are comparing channel interleaving between a dual channel and a quad channel set, both with single rank DIMMs. You will be interleaving between memory channels more with the quad channel, which can combat certain overhead, but with Zen and Zen+, you have to then use the farther memory channels, which adds to the latency (latency is roughly standardized accessing memory on Zen 2, so it is the same regardless of channel used, roughly, compared to prior gens). When you start using dual rank on dual channel, then use rank interleaving, you are now writing to four alternating ranks on the memory, with operations being able to access the other rank while the one just accessed is recovering. It is closer to what is seen with the quad channel memory with single rank, but without the near/far dynamic of earlier Zen and Zen+. But, the point is when the bandwidth is not so split and the cache is being kept fed by not splitting the bandwidth among the majority of cores, the performance benefit closes or shifts. On multi-tasking, depending on the task and memory needs, with high core count activation, there is without a doubt a benefit to extra memory bandwidth on specific workloads. This comes down, in part, to what the task is, how the software is optimized, and bandwidth per core.

I also was trying to explain the nuance of how mainstream systems with 8-cores and dual channel would have roughly the memory bandwidth per core of a 16-core system with 4-memory channels, but a quad-core with dual channel memory would have double the memory bandwidth per core, just like a 16-core with 8-memory channels would have. And that is where you got rude.

Now, as you can see, once a single core's memory bandwidth to keep the cache fed is satiated, that extra bandwidth doesn't mean ANYTHING. It is useless unless needed for a specific task. At that point, if the programs you use are not memory bandwidth limited, you would do better with taking less memory bandwidth with the mainstream platform dual channel, as many of those CPUs, especially on the Intel side, have higher boost speeds or can be overclocked higher, leading to more performance (single and lightly multi-threaded workloads benefit more from the frequency at that point than trying to get increased memory bandwidth). But because this is task specific and down to the programming of each program being evaluated, leaving it open ended in questions on which is better can create a situation where you get an "it depends" answer, something that is no help to anybody.

So I'm pretty sure on this one and gave a more nuanced answer with pictures to show that your instincts are correct for single threaded workloads (where the effects you spoke of would appear), but that for multi-tasking and heavier workloads, those worries can disappear in quite significant ways depending on the task.

There are a couple points that might make his findings easier to understand:
1) AMD's task balancing, specifically on streaming, sides towards devoting more CPU resources to the encoding/streaming side than Intel. This is why, in reviews of Zen and Zen+, AMD would often get hit worse on frame rates on the streamer side (actual game play) than Intel would. But, at certain loads, it could do better at not losing frames for the viewer side. I am assuming this has carried forward in some way to Zen 2. This behavior is also seen in some multi-tasking benchmarks for the release of Zen and Zen+, where for certain workloads TR would chew through both by allotting resources better for that multi-tasking, whereas other times Intel just laid the smackdown with multi-tasking.
2) AMD worked with Microsoft to change the scheduler behavior on AMD CPUs. This was to address the thread thrashing as well as thread propagation, making new threads use cores on the same CCX to lower the latency of cross-CCX communications, which is especially important since all inter-CCX communications now go through the I/O die. The scheduler changes would effect certain tasks, although I do not have a Zen 2 chip to check if streaming is one of the tasks benefited.
3) Some types of rendering are performed in cache, some from memory, some from I/O (meaning storage). Tile based rendering, for example, is done primarily from cache for some of it, which is something AMD excels at, hence good performance in things like V-Ray and Cinebench. Other rendering, it isn't as clear. If you look through Puget System's testing of the 3900X, there are times, even in Photoshop, that is now beats the Intel 9900K. They even recommend it in some cases for Premiere and Resolve over the Intel CPU.

This gets to the above discussion, in part, on bandwidth keeping the cache fed. There is a point on some programs where the issue isn't memory bandwidth, and in those cases the extra memory channels will not show a significant uplift in performance, instead favoring core speed and IPC. Which brings me to point 4.
4) AMD has a higher IPC than Intel's current chips. So at 4.15-4.2GHz, AMD's CPU is performing closer to Intel's chips at 4.4-4.6GHz. I don't know what he has his 7980XE overclocked to, if at all. But if it isn't overclocked at all, then the performance per thread would be higher on the 12-core than the 18-core. If it is, the single thread performance is roughly tied, but as mentioned in point 1, AMD devotes more resources to the encoding side. So if the performance has finally reached the threshold to not ding frame rates too heavily for the streamer, that really can be a boon for streaming. Further, with 6 fewer cores and 12 fewer threads, limiting OBS is less needed since it is sharing the resources with a game, meaning the game is taking enough threads to keep it from hitting the thread cap on OBS, thereby potentially reducing the setup time and the need to limit threads.This also would be related to the scheduler changes made in point 2.

I do agree, I would like to see some hard data on the topic from him and would love to see the head to head comparison to better know what he is seeing. He may do that closer to the 3950X launch, which I hope AMD seeds him with a chip for a review.

And I can agree on both the NVENC statement and wanting the head to head. I know he is knowledgeable, but also I am not knowledgeable enough to fully question his choices (hence why I like to have you chime in on the topic because I know you are WAYYYY more knowledgeable on streaming than I am).

 

Editing too difficult on phone atm, but two points to give. The first is thread limiting. The initial issue is that simply giving access to over x threads to the renderer starts negative scaling. It's like starting cinebench and superpi locked to a single thread affinity at the same time. The render thread in cinebench shared by superpi will simply be slower, it isn't ignored. Hence, if threadcount is the issue, "no tweaking" is a problem.

I watched it when tired last night so I didn't call this out but he mentioned both preview disabling and windows updates. This is something which I feel should cause a revisit to the 7980XE under the same conditions. Also yes I'd like to know what conditions his 7980XE was running under as well.

Sent from my OnePlus 6T using a bionic coconut

 

rofl https://www.phoronix.com/scan.php?page=article&item=sandy-extreme-2019&num=5

poor sandy-E getting destroyed.
7 yrs for intel to jump double in performance while took 2.5yrs for AMD to double. impressive.
give us that zen 3 already

 

Will this mean the shortages of mentioned chips will continue? Or will it affect their whole product stack of chips?

TSMC Trembles Under 7 nm Product Orders, Increases Delivery Lead Times Threefold - Could Hit AMD Product Availability techpowerup.com |Today, 17:01

TSMC is on the vanguard of chipset fabrication technology at this exact point in time - its 7 nm technology is the leading-edge of all large volume processes, and is being tapped by a number of companies for 7 nm silicon. One of its most relevant clients for our purposes, of course, is AMD - the company now enjoys a fabrication process lead over arch-rival Intel much due to its strategy of fabrication spin-off and becoming a fabless designer of chips. AMD's current product stack has made waves in the market by taking advantage of 7 nm's benefits, but it seems this may actually become a slight problem in the not so distant future.

TSMC has announced a threefold increase in its delivery lead times for 7 nm orders, from two months to nearly six months, which means that orders will now have to wait three times longer to be fulfilled than they once did. This means that current channel supplies and orders made after the decision from TSMC will take longer to materialize in actual silicon, which may lead to availability slumps should demand increase or maintain. AMD has its entire modern product stack built under the 7 nm process, so this could potentially affect both CPUs and GPUs from the company - and let's not forget AMD's Zen 3 and next-gen RDNA GPUs which are all being designed for the 7 nm+ process node. TSMC is expected to set aside further budget to expand capacity of its most advanced nodes, whilst accelerating investment on their N7+, N6, N5, and N3 nodes.

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



AMD's Ryzen 3000 Boost Fix Works, But Hits Wrong Cores tomshardware.com | September 17, 2019

 

So, let me put it this way: who bought the capacity through that period? It's a simple question. Everyone knows Apple uses the cutting edge node and takes up a significant capacity of that node, 25-50%, every fall. Since 7nm+ uses the same lines, once that contract kicked in, it was always going to squeeze capacity and extend lead times. The time period also coincides when a couple other key players were expected to use capacity in Q1, which is also done every year. But the real question is whether AMD increased their orders and contributed to the 6 months wait time. If AMD already had expanding production inked by now, which they would have an idea on necessary volume after seeing the 5700/XT popularity and the Ryzen 3000 and Rome chips, with AMD already keeping the yearly earnings forecast the same, but reduced Q3 forecasts which suggests increasing Q4 meaning their supply should be flush and flowing at that time, plus anything due for release at CES or within a month thereof.

So, although people are acting like doom and gloom on this story, they are forgetting AMD may be a significant cause of this situation, rather than impacted by it.

 

See, I put it together this way (quoted my old post) "AMD Ryzen 3700X and 3900X Shortages Still Persist Almost Two Months After Launch". I can't say this has an connection... But you see what I link together? There's a lot we don't know. But this does not look good <if> the "Shortages we have seen" is due to the report <from two months to nearly six months>I added into my post. If AMD has bought enough capacity... This does not exactly harmonize with the shortage of mentioned chips above. Could be AMD bin for server chips. Who know. But we both know that Apple was going to take a big slice of the cake. This was already mentioned by me many months ago.

See also from my older post from Aug 2018 AMD Is All In On 7nm CPUs For Strategic Advantage On Intel’s 10nm Stumble Hothardware.com

While it may sound like a gamble for AMD to put all of its eggs into a 7nm basket, it might be more out of necessity than choice. Intel contents that its 10nm process is more advanced than the competition, as it is packing more transistors and is doing a better job with scaling. In fact, Intel has even stated a desire to introduce a new metric to more accurately reflect where chip design technologies are really at...

 

This is the left over effect of GF abandoning 7nm. With the extra fabs in place shortages and other supply issue would have been reduced or eliminated. Now it is a severe demand issue and will probably get worse before it gets better.

 

charlie @ semi accurate

 

Yeah, exactly.


AMD Delays Ryzen 9 3950X Launch, Announces Third-Gen Threadripper With 24 Cores

AMD previously told us that it's long-awaited Ryzen 9 3950X, a 16-core 32-thread behemoth destined for the mainstream desktop, would arrive in September 2019, but today the company announced it is delaying the release until November while it focuses on meeting the demand for existing chips.

AMD's beastly 12-core 24-thread Ryzen 9 3900X is still a rarity at retail two and a half months after its launch, leading to price gouging from third-party retailers.

Yeah, No smoke without fire.

 

Wow, very disappointing to say the least. Two month delay on the 3950x and no telling when the 32 core TR will be around, and only an intent for the 24 core. While at the time AMD was hyping up putting TMC as exclusive 7nm supply I knew that would come back to bite them. They should have been at the ready when GF bailed out to find someone able to quickly put up a fab and get out silicon, if needed.

 

mmmmm i think there is more to that 24 core TR info than meets the eye. i am assuming they will make a more clear division between the regular TR and the WX TR skus this time around, thus the regular TRs will be bumped from 16 to 24 cores and the WX class from 32 to 48 cores. would make more sense than cutting the whole TR lineup at 24 cores, which is less than the 32 of the previous gen.

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

kinda sucks they dont allow OC for the WX version with 8 channel. maybe it is dual socket.

none of that matters though, we need these CPU run at minimum 4.8ghz, minimum 16 cores.

 

I do not think the TR limit will be 24 cores but I would have likened to see 32 cores released from the get go. Also the wait is almost unbearable. This as seeing the supply issues at present it looks like getting a high end Ryzen 3000 chip will be like trying to get a gold hens tooth.

Edit; the one save grace may be as TDP is 105w for the 3900x the power of the 24 core and heat should be manageable on most x399 boards.

 

As I pointed to Papusan, the yearly revenue forecast didn't change, just the quarter in which it is collected from third to fourth quarter. I believe that, said at Q2 earnings, is the delay being built in.

I think the mainstream 16-core delay is to work out the boost issue, especially since they saw more issues in speed with the 3900X than the 3600/3700/3800X lines. So that I think was unexpected, but a separate issue to supply.

 

If it is a hardware issue with boost speed any problem this month will still be a problem come November. So if this is the cause of delay it is a sever hit too Ryzen 3000. If it is a supply issue then it is just a temporary hit. My thoughts are with the 24 core TR is they are configurations that could not make the 32 core Epyc cut. Again this works with older x399 boards and their VRM's and cooling solutions etc. as if it replaces a x1950 it should just be a drop in and bios.

 

It is an algorithm and heat issue.

What is effecting the boost is part bios implementation of the AGESA (seen in hardware unboxed testing), part the move from 80C to 75C on temp to kick down speed to prolong longevity, part the high voltage levels implemented, etc. Most of it was solved in about a month, from it being late August to now for AGESA ABBA.

So, there is a potential to solve it by November, as the problem is known. With two extra months for a release, a lot can change. Look how much support was done in the first months of the first Ryzen release.

Aside from that, I think AMD locked down fab time weeks to months ago.

Finally, I disagree that 32 core is solely on the workstation board as that may have more pcie and ram, but it has no overclocking, etc.

Think of the Workstation platform, as currently envisioned, as a server chip to be thrown into a motherboard made for a tower. It's all the same.

Now, AMD released a 7371 Epyc that was a speed optimized 16-core. I believe AMD is looking at something like that for the threadripper workstation chips. But I also believe there will be up to a 32-core overclocking CPU, as that is needed to combat the 28 core Intel Xeon overclocking chip.

 

On the subject of Turbo Boost, Intel has discontinued Turbo Boost 3.0 driver, Intel recommends to uninstall it. Windows 10 Redstone 5 and beyond has the functionality built in, so you don't need the driver - but it might still be hanging around - if so uninstall it. The vulnerability is caused by the Turbo Boost 3.0 driver installer:

Summary:
"A potential security vulnerability in the Intel® Turbo Boost Max Technology 3.0 driver may allow escalation of privilege. Intel is not releasing updates to mitigate this potential vulnerability and has issued a Product Discontinuation notice for the Intel® Turbo Boost Max Technology 3.0 driver."

Recommendations:
"Intel has issued a Product Discontinuation notice for the Intel® Turbo Boost Max Technology 3.0 driver and recommends that users of the Intel® Turbo Boost Max Technology 3.0 driver uninstall it or discontinue use at their earliest convenience. This does not affect Intel® Turbo Boost Max Technology 3.0 functionality when used with in operating systems with native support (Windows 10 RS5 and later)."

Source: Turbo Boost Max Technology 3.0 Advisory
INTEL-SA-00243

I posted a bunch of Intel security patches / announcements a while back here's the rest in case you missed it:
http://forum.notebookreview.com/thr...ts-and-incidents.816109/page-25#post-10950408

 

AMD's 3rd Gen Ryzen Threadripper CPUs Rumored To End Support on X399 Chipset & TR4 Socket wccftech.com | Sept 27, 2019

AMD’s 3rd Generation Ryzen Threadripper processors are headed for launch in November. The new HEDT lineup would feature the brand new Zen 2 core architecture and tons of mega-tasking performance but along with all the added features, the processors might also require a new TR socket revision which if rumors hold true, might end their backward compatibility on the existing X399 products.

AMD’s 3rd Gen Ryzen Threadripper CPUs Rumored To End Support on X399 Chipset and TR4 Socket
The details come from the author of the DRAM Calculator for AMD Ryzen processors, Yuri Bubily or more famously known as 1usmus. He mentions on his post over at Overclock.net that AMD in their latest plans have removed X399 compatibility for next-generation Ryzen Threadripper processors. This clearly points towards the upcoming AMD 3rd Gen Ryzen Threadripper series which is going to make its debut with a 24 core and 48 thread chip in November. Following is the complete quote:

Hey. Unfortunately, there is not very good news, AMD has changed its mind about making X399 compatible with the new generation. For this reason, HEDT disappeared from all calendars and the release of these processors was forced to postpone.

Since the new processor has a new memory controller and he in a single copy (instead of two as before) had to seriously change the pins. Also, the new PCI Gen 4 standard and new power pins made a special contribution. TRX40 and TRX80 are a new generation and a new architecture. Nothing to do with the past generation. At the moment, all information is classified and I do not have access to it.

1usmus via Overclock.net​

 

Myself, I could easily see where the 48 and 64 core variants may need either the trx80 or wrx80 chipsets and thereby either a new socket or rework of the one at present for 8 channel memory. Now the post you refer too is back from 9/5/19, a bit old in the tooth and there are a lot of posts since.

Also they are now referring in that thread to the socket being the same but a different revision number, it is all rumors for now!

 

Ryzen 9 3900X Prices Have Gone Up Roughly 16 Percent Amidst Shortage tomshardware.com | Sept 28, 2019

Computer hardware, like any other consumer product, conforms to the law of supply and demand. What's pretty intriguing is that the Ryzen 9 3900X has gone up in price in a little more than three month's time since the chip's release, and though we've already seen extreme price gouging on eBay and other third-party sellers, now retail outlets have also raised prices.

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


Even at -180 Degrees, Ryzen 3000 May Not Hit Boost Clock Speeds tomshardware.com | Sept 28, 2019

As we've been covering, AMD's Ryzen 3000 series CPUs often have issues hitting and sustaining the advertised boost clock speed. A recent motherboard firmware update improved things a little bit, but often boosts the wrong core. So I set out to see if adding more cooling would help get my store-bought Ryzen 7 3700X to hit or exceed its advertised top-speed of 4.4 GHz. What I found out is that, even when I used liquid nitrogen to freeze my processor down to -180 degrees Celsius, it was still stuck at 4.35 GHz, 50 MHz below its boost. However, as we'll see below, you can hit boost clocks by manual overclocking.

 

At least with Newegg the chips are all outside vendors, price gouging. At this point there is a definite issue with supply and no end in sight. AMD is quickly falling into the realm of Intel with vapor wear territory here. Can't hit the clocks and then can't produce the supply of chips.

Now with November approaching it may not be until December or January before we see stores with a supply and by then we will be over halfway to the next launch, pathetic.

 

That's great news, more memory channels is a good thing and new memory channels can't be "downloaded" to upgrade and old motherboard.

It's a tough choice, but AMD will continue to sell and support ThreadRipper 1/2 CPU's and motherboards.

Adding a new motherboard generation now to get more memory channels is a tough choice, but I'm happy AMD did it.

Hopefully the pricing of the ThreadRipper TR5(?) CPU / motherboards won't clip us for much more than the TR4 generation, that's what I am concerned about right now.

It would be nice if we heard about TR5(?) before the 3950x releases, I'll wait to see the TR5(?) lineup CPU's and motherboards before deciding which to build for what purpose.

If the TR5(?) lineup looks good, then I'll have to wait before recommending that until I see the release results and still wait a good 3 to 6 months before building - to make sure it's got everything sorted.

TR580(?) and TR540(?), Exciting stuff.

 

Even in the same article they are quoting the 9/5 reddit post. At the same time they admit it is just a revision of the same socket and might allow old chipsets but with out all the feature set. I would not jump in so fast with the doom and gloom for x399 owners yet.

 

It might be possible to use the new CPU's with reduced memory channels in the old motherboard - if it's the same socket and those same pins aren't used for something else, but that would have required planning it out from the start - and I don't think that's been anticipated.

I recall when AMD moved from DDR2 to DDR3 and you could put the same CPU in 2 different motherboards with support for either, but AMD telegraphed that feature well ahead of time.

Of course we don't know, this could have been the plan all along and if so then there could be support for current TR4 motherboards.

But, unlike the PCIE4.0 upgrade a memory upgrade - more channels with better performance in more applications - all applications if done right - would be more of a reason to upgrade to the new motherboard with more channels.

Given that scenerio AMD might decide to forgo disappointing current owners with the comparison between their purchased CPU in their previous generation motherboard vs reviews / others results in the new motherboards.

Tough choice, but sometimes you have to lead the industry rather than coddling to it.

No wonder AMD hushed up on ThreadRipper and have been quiet except for confirming the line will continue. Too much to explain and describe in understandable form - their reasoning and the results - that would have been too distracting during the Ryzen 3 roll-out and would still be distracting up through the 3950x (3950 too?).

I wonder if the ThreadRipper line might slide forward like the APU's being 1 generation lag behind Ryzen and Epyc moving forward?

 

TR4 is supposed to be supported too 2020 and I can see where up to 32 cores would be ok as they can do that now. With the TR80 and WRX80 I can see where that is needed for 48 and 64 cores. I believe this is the reason for the 24 core release.

1.) power profile and TDP should not be way over the original 180 watt.
2.) Quad channel memory should be no more a constraint than dual channel on the 3900x
3.) PCI-e 3.0 will not be as much a constraint.
4. Easier for a CCX configuration to be compatible with x399.

Etc., Etc., Etc.. Again do not count out x399, but maybe after this it will be dead in the water. I assume this may be true in the future of AM4 as well.

 

They do the one gen behind, no one will buy them. APUs are a LOW end product, comparatively. HEDT is HIGH END. If it lags a generation EVER, consider them opening the lane for Intel to take it back. Not going to happen. Also, it doesn't make sense for numerous other reasons. But it always comes out after Ryzen and Epyc to get the binned chips. So there will be a lag. But generational lag, no!

Yeah, either they are going to go to all exclusive 8 channel memory and the traces don't exist for X399 support and they can't figure out pinning or not.

Here's what likely will happen:

1) an X399 compatible tr4 socket 24 and 32 core chips with only quad channel support (32 core is delayed because due to the boost issues, primarily due to voltages with mb MFRS and the algorithm issues, as those boost speeds can be reached through manual overclocking, they are taking the cherry 8 core chiplets to now use on the 3950X, which is why it is now delayed until TR launch and the 32 core disappears with the 16-core mainstream in it's place).

2) the WR8 socket, or whatever it is called, has 8 memory channels supporting speed optimized 48 and 64 core Epyc CPUs. Straight forward. If you need more than 32 cores, buy the Epyc workstation platform.

3) the TR8 motherboard and chips are for the Zen 3 based CPUs and will support the entire lineup. This will not be here until next year. It doesn't apply to Zen 2 at all. It will not support quad channel memory chips, will have no backwards compatibility, and the chips designed for it will not be compatible with X399 at all. It may be compatible with Epyc CPUs as well as the speed optimized Epyc CPUs. That depends on whether they put 64 or 128 PCIe lanes on the TR8 motherboard. The workstation version will have all 128 lanes, but they are locking down overclocking.

That makes the most sense of all. Not chip shortages causing the 3950X delay, blah, blah, blah.

The only question is whether they worked out the TR8 socket to have more than one generation of compatibility to support Zen 4 or beyond. They may even use it as a test for the new socket for Epyc coming in 2021. But that is my speculation. Why? Because that makes the most sense.

 

As it stands I can not use all of my 1950x as is. A 24 core would be perfect as it should produce close to 3900x performance for single core and close to 2x performance over the 1950x in CB-R15 over stock (if it scales 100% over the 3900x). While yes 32 or even 48 and 64 cores sounds nice at some point reduced clock rates will provide diminishing returns.

 

Agreed. I was hoping they would do a 16-core TR for entry as dual channel memory isn't going to happen, nor dicing my PCIe lanes. Now I just have to hope they binned the 6-core chiplets well and hope for a price reduction from the prior generation.

 

Highlights of the day: Tight supply at TSMC, not just for 7nm Digitimes.com | 28, 2019

------------------------------- Still, None know how AMD is into this.

AMD’s 12 Core Ryzen 9 3900X CPU Shortages Prevails, Selling For Up To $800 US at Major Retailers, Not Available on Others wccftech.com | Sept 29, 2019

 

Sold and shipped by "Deals a Day" so it is outside vendors price gouging.

 

It's interesting, this article ends with:

"In addition, AMD has never promised a certain longevity or compatibility of the socket TR4 for future generations."

Is that true? I've seen AMD say 4 generations (or 4 years) on AM4, but no such statement for TR4...?

" Mark Papermaster, AMD’s chief technology officer, confirmed the four-year lifespan in a conversation here at CES 2017 in Las Vegas, though he declined to discuss specifics."
https://www.pcworld.com/article/315...hitecture-is-expected-to-last-four-years.html

3rd generation of Ryzen Threadripper apparently requires new motherboards [German]
Thursday, September 26, 2019 at 17:38 by Andreas Schilling
https://www.hardwareluxx.de/index.p...er-setzt-offenbar-neue-mainboards-voraus.html

 

From the article," Zen to live on
While Anderson’s responsible for bringing Ryzen to market—“you don’t have any idea how many hours I and my team have spent on this,” Anderson said—it’s Papermaster who has to think of the future. When asked how long Zen would last, compared to Intel’s two-year tick-tock cadence, Papermaster confirmed the four-year lifespan and tapped the table in front of him: “We’re not going tick-tock,” he said. “Zen is going to be tock, tock, tock.” "

So Zen was to be four years of tock, tock, tock, not AM4. TR4, or Epyc socket.

So TBH, I will be quite bent out of shape if they did not announce as soon as they knew they would have to abandon x399, Intel was at least forth right letting people know they needed new chipsets early on. If AMD is going to go back on its word they should have been forthright to begin with also. How many people bought into a TR 2000 not knowing they were getting and end of system upgrade.

 

Well we really don't have AMD's official words on this yet, it's all extrapolation based on hearsay and rumors, so we wait till November to find out for sure when AMD announces what ThreadRipper 4 will be all about - socket(s), motherboard(s) supported, CPU's, and what the migration plan is there if any from TR4 1st/2nd generations to TR4(?) 3rd generation.

 

You guys are hitting on why my above statement of sTR4, sTR8, and the 8 channel workstation makes sense.

They would not release the 24-core in November without motherboards and we have seen ZERO leaks on motherboards, other than the registration of the 8-channel boards, to date. We are a month or so before those boards would need ready for a 24-core, yet we see nothing. We've already got B550 boards being shown for APUs and lower end Ryzen chips coming soon. Still we have nothing on new HEDT boards. The boards shown off at CES were a month or so before the launch of Ryzen 3000 in July.

Because of this, I'm thinking we will have the two tier system this year, X399 and the workstation boards (which don't always get shown before release as they are similar to servers in that way/don't need hype).

Moreover, if Zen 3 is getting new 8-channel boards in 2020, which red gaming tech's original leak said we would only get the workstation boards and 4-channel TR4 boards this year, combined with AMD's prior statements of compatibility UNTIL (not through) 2020, even though often in the context of AM4 and Epyc, by waiting until then for the new TR8 boards, they will not have lied to the market, even if it is questionable whether they ever meant the statement to apply to the HEDT segment.

I believe they are also making the move to 8-channels due to professionals asking for memory bandwidth (as well as it becoming more obvious that it is needed).

So, until the rumors are confirmed, and because we've seen nothing of motherboards, actual motherboards, this close to launch, people should remain calm and wait.

Not only that, if you look at earlier rumors, new boards were said to be released around CES, as well as not all of the TR lineup being released this year, with some coming early next year.

Take those all together and it looks like people are being hysterical when all the leaks that seem credible, taken together, suggest what I said in my above post and here. There was going to be a 32-core this year until the 16-core boost issues arose. There is solid reasons to think they would then push off the 32-core for both Epyc demand and to not overly delay the 3950X, while getting 2x 3950X for every 32-core chip they would have produced, thereby also giving a more ample volume to the market right in time for Christmas.

 

I am stating that a single post, by a non English speaker, on 9/5 is being taken as gospel. No corroboration etc.. So my point is wait and see but if it were true I would be very upset.

 

To be clear, I was not meaning you two were being hysterical, rather all of the coverage and other people in forums, to varying degrees, are being such. Not only that, 1usmus clearly stated that he has no access to the documents and that the documents are under NDA. Although he has worked with AMD in the past, that means all information he has is second hand. This is where it can be easy to mix up which generation this will be applied to and when it is happening.

Moreover, the other rumors, such as the initial rumors from red gaining tech, came from sources inside the industry that he cited the prior leaks covered in which the information provided was valid.

I do apologize if it sounded like I was accusing you two specifically of hysteria.

 

https://www.tomshardware.com/news/amd-threadripper-3000-castle-peak-32-core-cpu,40151.html

Above shows the 32 core Sharkstooth running on a Whitehaven board or first gen. TR. So AMD an easily run up to 32 cores on x399.

 

To show a point. I am running my 1950x at 4.0 GHz on 1.350 vcore on my x399 Asrock Taichi. I just ran 10 rapid CBR20 runs on it and CPU never went over 74c and VRM never over 51c. Actually it got to the max temps on the 5th successive run but I went too 10 just to be sure. Also note the package went to 287 Watts.

So this board should easily handle 32 core Zen 3 stock and maybe even slightly overclocked.

Edit; I should also note if they scale 100% over the 3900x the 32 core TR may spank a dual Intel 8180 stock in both CB R15 and R20.

 

https://videocardz.com/82105/intel-...cascade-lake-x-final-specs-and-pricing-leaked

 

Interesting, I'm glad Intel drastically reduced their prices. Who knows, maybe even I'll build a cascade lake x box for fun.

 

And with cutting the 18-core in half, added to Intel already stating 14nm supply issues will return in Q4, Intel is really struggling.

And with the quality of AMD chips and the rise in servers, AMD is also having supply issues to a degree due to demand.

So other than demand issues with both companies, the truth is this is flipping amazing for core counts so low.

Notice Intel will not sell a 16-core. I think they don't want a head to head Intel versus AMD, which reeks of fear.

At 4 more pcie on platform, still less than AMD, along with prior gen. Only matters if you can use them, though.

Intel is going to struggle until 2021 without a doubt.

Also, 7xxx and 9xxx chips are going to be SOOO cheap moving forward. If performance doesn't increase much with the 10 series, picking up a high core count for cheap used seems like a really good deal for those wanting Intel. We'll see in November for both companies.

But this is better than I thought on pricing. Kudos. The 10940X is positioned to go against the 3950X. That will be a real interesting match up!

Now where does AMD price the 24 core TR? With those prices, I almost want to bet $1200, but am sure they are checking the numbers right now. Any lower = heaven.