Separate names with a comma.
Discussion in 'CPUs and Overclocking' started by CommManagerJ, May 7, 2018.
No, they are currently only for system builders and not sold individually.
To ensure our resources are focused according to business priorities, the documentation is available to AMD partners under specific NDA access.
As a C++ programmer and an AMD fan, I'm always looking for AMD participation at C++ conferences. Does AMD actively participate in the C++ standardisation and language evolution process? C++ has been rejuvenating the last decade and is increasingly the go-to language for well optimised and power-efficient code in all areas from embedded, mobile and desktop to server.
PS. You should send a cake and congratulate your partner Microsoft for finally achieving C++ standard conformance in MSVC.
Per the agreement in the header, all queries may not be answered.
I have a question about cache latencies: It appears I own a 1600X with 12-cycle L2 latency, and "enhanced" L3 latency. (I have not tested AGESA before SummitPI 220.127.116.11). The CPU was manufactured in early 2017, and purchased last June.
As far as you know, were there any Summit Ridge dies that released with enhanced cache latency, or maybe a change in AGESA configuration allowed for better cache latency on certain models?
At stock speeds, here are my AIDA results. (The 4041 MHz is detected XFR clock @ 40.5 multiplier)
The 3.1ns result is consistent with what is quoted for Pinnacle Ridge, not Summit Ridge. I asked Robert Hallock earlier today, and he seemed a little surprised by my results. The 10ns L3 result also seems to be improved.
At 4 GHz (OC), the results for L2 and L3 cache are 3.0ns, and 10.0ns. This is consistent, and I am sure it is not an error. At 4 GHz, 3.0ns is exactly 12 cycles (0.25ns * x cycles = 3ns) and the 10ns equating to 40 cycles (0.25ns * x cycles = 10ns)
Upon release, Nathan Kirsch at LegitReviews had results more consistent with what you guys believe should be the norm for Summit Ridge: 4.2ns L2 latency (~17 cycles), and 11.1ns latency (~43 cycles).
Any reason why I'm seeing this increased performance? Some others with different Summit Ridge SKUs (1600, 1700, 1800X) I have seen recently still have ~17 / 43 cycle L2/L3 latency results.
@BeepBeep2 on some X370 boards there is a "Performance Bias" option that will allow you to run more aggressive settings, and I believe that is why you are seeing better #s:
*Some* Ryzen 1000 series chips are capable of running stably at those settings. All Threadripper chips had the reduced latency out of the box.
Interesting. I'll go test with Performance Bias off. So far, I only tested with CB11.5, CB15, and AIDA modes.
A user at chiphell on SummitPI 18.104.22.168 had similar results on an 1800X with the same motherboard.
Thanks! My performance is now worse with performance bias off. On AUTO, it is still lowering latency.
Thanks for the AMA. My question is, what is in store for AGESA updates? Now that Pinnacle Ridge is out, it feels like Summit Ridge, and Raven Ridge seem a bit left out. Are the Bios updates/AGESA updates supposed to be updated so that every platform is on one unified BIOS, regardless of CPU used? And how closely do you work with the motherboard manufacturers to ensure quality and stability of what's being released?
So no comment/reply on efforts to get embedded products out to interested enthusiasts from post 76? Thanks for any responses....mostly wondering if my post was missed or skipped...very interested in epyc 3000 for home server and v1000 tiny htpc.
As an owner of a 1900X are there certain sub timings in RAM that manual tuning would benefit Threadripper performance?
(I’m running 4X8GB 3200mhz GSkill Samsung B-die 14-14-14-34-1T)
Edit: AND I should have read the entire thread. Almost a whole page dedicated to timings.
I have a question about Ryzen laptops. So far, the reception is a bit hit and miss. On the good side is that the iGPU beats any Intel iGPU by a large margin and also the CPU performance is quite good. On the bad side, the main criticism is that battery life is poor compared with equivalent Intel laptop when doing light workloads like web browsing or watching videos. Is AMD working with the OEM to try to resolve these issues? I mean, the problem could be the components the OEM has chosen, or it could be BIOS or drivers need optimisation.
Some of us build computers... admittedly much smaller volume than any of the OEM, but we'd like to have the same parts available.
I've built more than a few custom systems for aerial photography, had to run off of a cigarette adapter, so max of like 50W, total system draw.
Of course I can always do what I did then and use a desktop chip, undervolted and underclocked.
There have been a lot of AMD systems taking aerial photos because they're unlocked and I can use a true desktop chip. Jaguar and Atom don't do well with 100MB per pic @ 1 pic per second.
My questions will be about Raven Ridge capabilities and also about tuning of these APUs to become the best products in the market:
1) When testing laptops with Raven Ridge, monitoring software sees DDR4 to switch memory clock 1866 to 2133 to 2400 (933-1066-1200MHz). Is this a new technology on AMD's part (presumably to save power)- or a fault in monitoring software/data reporting? Only seen with laptop Raven Ridge.
2) In laptops, power and it's balancing is extremely important. For example, on Acer Aspire3 with Ryzen 3 2200U, pulling the power plug out (or setting Maximum Processor State to just 50%) does not allow CPU clocks to boost over 1600MHz, but saved power is then redirected to boosting iGPU, significantly increasing iGPU clock and game performance. Maybe AMD could consider allowing users to adjust power distribution between the CPU and iGPU manually, preferably in game, through Radeon Overlay, to get the best performance?
3) It is nice to see many new laptops coming soon, but so far- no Raven Ridge laptops got Freesync. Could you comment- is it an expensive feature to implement with low (30-60Hz) frequency range, which is most beneficial for such iGPUs? Should we be demanding laptop manufacturers to include it?
4) Intel has created succesful 'Centrino', 'Ultrabook' platforms with qualifying requirements. Maybe AMD could benefit from their own platform- like strict requirement for laptop/AIO having: an SSD, IPS with Freesync, 25W+ cooling, 2300U or better, and dual channel RAM? These are all very reasonable requirements, achievable even for a budget laptop. This would bring some decent quality standard- and allow to easily recommend- 'Get this xxxxx AMD platform for good components and value!'.
The AGESA updates are in fact converged to cover all products supported on the platform. And while the focus is currently on the new 2nd Gen processors, there are updates inside the code we provide to the motherboard manufacturers that improve all products.
We work very closely with the manufacturers, a dedicated team in both the US and in Taiwan engages with the motherboard BIOS teams to ensure the updates are understood, and any issues or questions addressed. We don't dictate to them schedule or versions but instead work together to establish priorities appropriately for the goals of our combined businesses, with a strong focus on customer feedback.
Yes, absolutely we're working with the manufacturers. We don't want bad experiences for users, and neither do they.
1 - Yes, this is an AMD feature to help save battery power by adjusting memory speed dynamically. It's pretty nifty!
2 - Typically the manufacturers do not want performance adjustments at a silicon or firmware level performed by the end user. But I'll pass along the feedback and see if there's an option for us to look at there.
3 - If you want Freesync available, absolutely you should ask the manufacturers of notebooks to offer it as an option.
4 - Those programs are typically incentivized in some manner, and I'm not sure that those incentives align to the company growth goals right now.
Thank you for the response. I was very curious about this, as some board partners are hit and miss with support and what's working with Ryzen Gen 1. I have a 2nd gen Ryzen along with a 1st gen, and the 1st gen does not play too well with the Pinnacle Pi BIOS'. Hopefully we'll see some updated support that'll work besides partners telling us to downgrade back to 22.214.171.124b, etc.
I don't know... sorry. I would imagine we have working group participation at many levels but dont know for sure.
As we enjoy more success, it will get easier to be able to support the broader reaches of the market. I'm sure at some point they'll appear in pre-built PCs from local system builders.
Pretty sure DTX is dead dude... sorry
There's a lot of opportunities in front of AMD right now, which means not every niche will get filled instantly. Given time, we'll address all the markets and find ways to get all our tech into the hands of the people who want it.
If double dipping is okay, I have another question.
1. This is about your potentially future value ULV or low power, sub-15W options, akin to Intel’s Atom and the non-Core-based Pentium product lines. A couple years ago, rumor and leaks had it that a low power line of processors, codenamed Banded Kestrel, was in the pipeline. Now, in spite of Intel Atom’s well-known failure in mobile, fragments and derivatives of it still live on. The resulting market thrives in the form of low cost laptops, tablets, and 2-in-1’s from the likes of Chuwi, Teclast and other far eastern outfits as well as more specialized designs such as the GPD Win and GPD Pocket. If you cannot say anything else, does AMD still plan on addressing this market segment? Beyond that, if so, will it be with Banded Kestrel? Or perhaps does it instead plan on doing so with products using cut down Raven Ridge dies to serve the same purpose?
Thanks a bunch....kinda assumed this stuff was a lower priority.
On a similar front...has there been any discussion of offering a single die AM4 based Epyc to go after Xeon E3...not sure such a thing is a necessary product on Intel side with Xeon D...but maybe makes more sense with AMD sockets remaining compatible over larger time-frames and more product cycles than Intel.
Good day James,
I'd like to thank AMD for producing Threadripper and for enabling ECC on it. My Threadripper 1950X is an awesome CPU indeed. I'd like to take this opportunity to suggest some features that I would love to see in the future.
1) RDIMM/LRDIMM support that would make it possible for users to reach (or at least get much closer to) Threadripper's advertised maximum of 1TB, which as of now is woefully inaccessible with support being limited to UDIMMs.
2) A dual configuration of Threadripper would be very welcome.
3) A fully enabled, 32-core Threadripper with all dies populated (instead of the current 2 dies) that is made with the absolute top binned dies available would be an amazing chip and many people (myself included) would gladly pay a substantial premium for such a chip.
Thanks again for producing Threadripper...I hope to see it made even better in the future!
Thanks for doing this, James!
I have a few of questions:
1. There are rumors that Zen has an amazing wafer success rate. Any comment on Zen's prowess in the fab?
2. Any lessons learned from the issue of Raven Ridge compatibility on older AM4 motherboards? Does AMD plan on changing their chipset release strategy to allow for a better out-of-the-box experience for such CPUs? For example, Raven Ridge compatibility probably wouldn't have been as messy had first-time Ryzen buyers been able to buy a new B450 motherboard rather than a B350 with an older BIOS.
3. How much does the fabrication technology influence the architecture of your processors. For example, is Zen 2 designed specifically with Global Foundries' 7nm process in mind? Since you use TSMC for the the custom SOCs used inside Playstation and Xbox, are architecture tweaks a necessary component of working with their fab?
4. This one is probably more Radeon-related, but what are your thoughts on providing a dedicated graphics cache for future Zen APUs? Adding something like the Xbox One's 32MB of ESRAM might dramatically improve memory bandwidth for future Vega/Navi GPUs that come in the APU lineup.