HW News - AMD Delays 3950X w/ Threadripper, 7nm Supply Hurting, R5 3500X
video description
Date: 2020-05-06
Related videos
Comments and reviews: 10
ajc9988
So, on the interconnect story, the way you worded the ending is wrong. CXL supplants CCIX. That is absolutely correct. But CXL uses PCIe 5. 0 PHY, CCIX was initially designed for PCIe 4. 0 PHY. Moreover, the incorrect part came at unclear language at the end suggesting that CXL wins against Gen Z. In fact, those are planned to be used in conjunction, with CXL handling everything inside a node, while Gen Z is used to connect nodes together. Think of CXL as the coherent interconnect for the MB, CPU, and components normally connected to a system while Gen Z is taking the place of connecting two systems together to act coherently together. Gen Z is closer to like Slingshot being used by Cray for the Frontier (and Shasta, if memory serves) supercomputer(s. The interviews with HP and Dell executives, two of the largest OEMs/system integrators, are what solidified the use differences for me a couple months back. The are committed to using both in upcoming systems, which with the split like that, intranode and internode, it makes sense as you are not compromising on features needed but lacking for each of those purposes, which would happen if you only used one standard. But that is my understanding and I welcome being proved wrong.
reply
So, on the interconnect story, the way you worded the ending is wrong. CXL supplants CCIX. That is absolutely correct. But CXL uses PCIe 5. 0 PHY, CCIX was initially designed for PCIe 4. 0 PHY. Moreover, the incorrect part came at unclear language at the end suggesting that CXL wins against Gen Z. In fact, those are planned to be used in conjunction, with CXL handling everything inside a node, while Gen Z is used to connect nodes together. Think of CXL as the coherent interconnect for the MB, CPU, and components normally connected to a system while Gen Z is taking the place of connecting two systems together to act coherently together. Gen Z is closer to like Slingshot being used by Cray for the Frontier (and Shasta, if memory serves) supercomputer(s. The interviews with HP and Dell executives, two of the largest OEMs/system integrators, are what solidified the use differences for me a couple months back. The are committed to using both in upcoming systems, which with the split like that, intranode and internode, it makes sense as you are not compromising on features needed but lacking for each of those purposes, which would happen if you only used one standard. But that is my understanding and I welcome being proved wrong.
reply
Hairy
The Windows scheduler thing is also why I think Der Ba8er's survey has a problem. AMD only guarantees the boost clock on one core, the one marked with the gold star in Ryzen Master. When I run the CB single core test on my Ryzen 3600 it does not get scheduled to that core, but to a random core. In fact, in spite of the name, I see loads peak on different threads during the run and I am pretty sure not all are on the same core (I use CB20 instead of CB15) And there is no way to easily force Windows to use the gold star core (Maybe it can be done with Process Lasso) If this is the case, it is no wonder that in Der Ba8er's survey the number of Ryzen 3000 CPU's that can't reach their advertised boost clock gets worse with a rising number of cores. The chance that the bench mark is run on the gold star core just gets lower. Of course, this is not the entire story, because AMD's ABBA Agesa revision did help. But I still think Windows not assigning the CB single core benchmark run to the right, 'gold star' core plays an important role in Der Ba8er's survey results.
reply
The Windows scheduler thing is also why I think Der Ba8er's survey has a problem. AMD only guarantees the boost clock on one core, the one marked with the gold star in Ryzen Master. When I run the CB single core test on my Ryzen 3600 it does not get scheduled to that core, but to a random core. In fact, in spite of the name, I see loads peak on different threads during the run and I am pretty sure not all are on the same core (I use CB20 instead of CB15) And there is no way to easily force Windows to use the gold star core (Maybe it can be done with Process Lasso) If this is the case, it is no wonder that in Der Ba8er's survey the number of Ryzen 3000 CPU's that can't reach their advertised boost clock gets worse with a rising number of cores. The chance that the bench mark is run on the gold star core just gets lower. Of course, this is not the entire story, because AMD's ABBA Agesa revision did help. But I still think Windows not assigning the CB single core benchmark run to the right, 'gold star' core plays an important role in Der Ba8er's survey results.
reply
Anthony
AMD didn't offered some precision about November and as we can see for the 3900x availability, it won't never be really present on retailers even at it November launching. 3950x is only a 2019 P A P E R L A U N C H. it won't never be available before Christmas at least, so maybe January or more. it will be interesting to see the behaviour of its customers target (AMD Enthousiasm) with its true availability and the approaching Zen3 upgrade. not sure they won't pay so much / when the next gen is in less than 6 months. And no one can know if they won't delay a new time in November because of any reason (TSMC, good 8c chiplets more focused for Threadripper which is more profitable in terms of margin, etc. Next time AMD, please don't lie and only talk about true availability for Customer's respect. the contract between AMD and TSMC, it is not a simple handshake between a grocery store and a farmer on an estimation of eggs that will produce the hens in 1 month.
reply
AMD didn't offered some precision about November and as we can see for the 3900x availability, it won't never be really present on retailers even at it November launching. 3950x is only a 2019 P A P E R L A U N C H. it won't never be available before Christmas at least, so maybe January or more. it will be interesting to see the behaviour of its customers target (AMD Enthousiasm) with its true availability and the approaching Zen3 upgrade. not sure they won't pay so much / when the next gen is in less than 6 months. And no one can know if they won't delay a new time in November because of any reason (TSMC, good 8c chiplets more focused for Threadripper which is more profitable in terms of margin, etc. Next time AMD, please don't lie and only talk about true availability for Customer's respect. the contract between AMD and TSMC, it is not a simple handshake between a grocery store and a farmer on an estimation of eggs that will produce the hens in 1 month.
reply
Kamil
Next month i will upgrade my gamin pc and i need some help. What is the fastest processor for gaming right now? Im thinking intel 9900k but im not decide yet its not that new anymore. Buy it now or wait for intel or amd new gen? I dont upgrade pc often so i need somethink really good for about 3 years. My first upgrade will be processor(old 6700k right now) motherboard( old z170 msi) and ram( old corsair dominator 16gb 3000mhz cl16) and maybe some watercool components, my graphic cards will stay with me for next gen come out, becouse those new rtx are overpriced in my opinion and not that crazy fast as they say, compare to older gtx gen i know new shadowa or somethink but while playing i must stop game to see real deference so maybe for a while.
reply
Next month i will upgrade my gamin pc and i need some help. What is the fastest processor for gaming right now? Im thinking intel 9900k but im not decide yet its not that new anymore. Buy it now or wait for intel or amd new gen? I dont upgrade pc often so i need somethink really good for about 3 years. My first upgrade will be processor(old 6700k right now) motherboard( old z170 msi) and ram( old corsair dominator 16gb 3000mhz cl16) and maybe some watercool components, my graphic cards will stay with me for next gen come out, becouse those new rtx are overpriced in my opinion and not that crazy fast as they say, compare to older gtx gen i know new shadowa or somethink but while playing i must stop game to see real deference so maybe for a while.
reply
K31TH3R
The Windows Scheduler, surprisingly, is always targeting one of the correct fast cores of my 3600X, but not the fastest core. Because of this, on ABBA, I've seen 4490MHz on that core a few times, but usually 4417. My fastest core is core 5, which I don't ever see going over 4342MHz. However, with PBO enabled, I'm now seeing all cores boost up to 4342MHz simultaneously in light multithreaded workloads. This is a better result of my all core 4315MHz overclock at 1. 36v. So now it seems the Windows scheduler is the last hurdle to overcome to seeing the full potential of Zen2.
reply
The Windows Scheduler, surprisingly, is always targeting one of the correct fast cores of my 3600X, but not the fastest core. Because of this, on ABBA, I've seen 4490MHz on that core a few times, but usually 4417. My fastest core is core 5, which I don't ever see going over 4342MHz. However, with PBO enabled, I'm now seeing all cores boost up to 4342MHz simultaneously in light multithreaded workloads. This is a better result of my all core 4315MHz overclock at 1. 36v. So now it seems the Windows scheduler is the last hurdle to overcome to seeing the full potential of Zen2.
reply
Nakatomi
Corsairs iCue software is awful I have a pair of Void wireless RGB 7. 1 SS Headset and I have to use the original software that came with it 2. 2. because the latest of iCue causes more issues than original software, it causes lock ups and sometimes won't let me switch between stereo and surround sound while in game. And as you said Corsair take forever to even fix anything and they don't always fix the issues on there own forum and sometimes it makes new bugs
reply
Corsairs iCue software is awful I have a pair of Void wireless RGB 7. 1 SS Headset and I have to use the original software that came with it 2. 2. because the latest of iCue causes more issues than original software, it causes lock ups and sometimes won't let me switch between stereo and surround sound while in game. And as you said Corsair take forever to even fix anything and they don't always fix the issues on there own forum and sometimes it makes new bugs
reply
Mr.
Not surprised. they can t even get the 3900x out. I am stunned AMD actually told people it s delayed because their communication has been garbage. I have the board, RAM, all I need is the CPU. Past the time to return crap or I would have stayed Intel, unless this chip is blow my mind life changing. Next upgrade will be back to Intel, they actually release what they say. Hit their claimed speeds, no bios issues, no real heat issues. 3900x release has been a joke.
reply
Not surprised. they can t even get the 3900x out. I am stunned AMD actually told people it s delayed because their communication has been garbage. I have the board, RAM, all I need is the CPU. Past the time to return crap or I would have stayed Intel, unless this chip is blow my mind life changing. Next upgrade will be back to Intel, they actually release what they say. Hit their claimed speeds, no bios issues, no real heat issues. 3900x release has been a joke.
reply
Jeffrey
OVERPOLLING (or some other USB issue) still happens in the older game F. E. A. R.; I forget the details now but I just fired it up to check and the FPS was way lower than it should be. Something like 25% of expected or lower. The solution I forget but it involves unplugging some USB peripherals I believe. PCGAMINGWIKI's FEAR page mentions Logitech devices and there's some sort of patch linked. sad to see this issue can still happen in 2019.
reply
OVERPOLLING (or some other USB issue) still happens in the older game F. E. A. R.; I forget the details now but I just fired it up to check and the FPS was way lower than it should be. Something like 25% of expected or lower. The solution I forget but it involves unplugging some USB peripherals I believe. PCGAMINGWIKI's FEAR page mentions Logitech devices and there's some sort of patch linked. sad to see this issue can still happen in 2019.
reply
supra107
I have a Corsair keyboard and a Logitech mouse, and comparing Logitech's software to Corsair's software, Logitech's soft is incredibly optimized compared to Corsair's. iCUE eats up around 400000K of RAM, while G HUB in total eats up around 200000K of RAM. Yes, it's still a lot, but compared to iCUE or Chromium it's much better. Also, it's much more useful than iCUE, since I find mouse macros to be more versatile than keyboard macros.
reply
I have a Corsair keyboard and a Logitech mouse, and comparing Logitech's software to Corsair's software, Logitech's soft is incredibly optimized compared to Corsair's. iCUE eats up around 400000K of RAM, while G HUB in total eats up around 200000K of RAM. Yes, it's still a lot, but compared to iCUE or Chromium it's much better. Also, it's much more useful than iCUE, since I find mouse macros to be more versatile than keyboard macros.
reply
Tim
I have to turn off ICUE in order to play Wireless VR, other wise even with Steam VR home turned off I m hitting 80-90% MT usage on my 7700k. The only reason I found out was do to adjusting my memory Timings to try and help my FPS in VR, and in my testing I seen my CPU OC to 5. 1GHz was getting 400pts less in CineBench then my CPU running stock. So with trial and error I now now to turn off ICUE in order to play VR now.
reply
I have to turn off ICUE in order to play Wireless VR, other wise even with Steam VR home turned off I m hitting 80-90% MT usage on my 7700k. The only reason I found out was do to adjusting my memory Timings to try and help my FPS in VR, and in my testing I seen my CPU OC to 5. 1GHz was getting 400pts less in CineBench then my CPU running stock. So with trial and error I now now to turn off ICUE in order to play VR now.
reply
Add a review, comment
Other channel videos
