Ripping threads with Coreprio and DLM

Subject: Processors | March 20, 2019 - 04:34 PM |
Tagged: amd, coreprio, threadripper 2, 2990wx, dynamic local mode

Owning a Threadripper is not boring, the new architecture offers a variety of interesting challenges to keep your attention.  One of these features is the lack of direct memory access for two of the dies, which can cause some performance issues and was at least partially addressed by introducing Dynamic Local Mode into Ryzen Master.  On Windows boxes, enabling that feature ensures your hardest working cores have direct memory access, on Linux systems the problem simply doesn't exist.  Another choice is Coreprio, developed by Bitsum, which accomplishes the same task but without the extras included in Ryzen Master. 

Techgage ran a series of benchmarks comparing the differences in performance between the default setting, DLM and Coreprio.

AMD-Ryzen-Threadripper-2990WX-CCX-Diagram.png

"Performance regression issues in Windows on AMD’s top-end Ryzen Threadripper CPUs haven’t gone unnoticed by those who own them, and six months after launch, the issues remain. Fortunately, there’s a new tool making the rounds that can help smooth out those regressions. We’re taking an initial look."

Here are some more Processor articles from around the web:

Processors

Source: Techgage

AMD States Its CPUs Are Not Susceptible to SPOILER

Subject: Processors | March 18, 2019 - 08:38 AM |
Tagged: spoiler, speculation, spectre, rowhammer, meltdown, amd

AMD has issued a support article stating that its CPUs are not susceptible to the recently disclosed SPOILER vulnerability. Support Article PA-240 confirms initial beliefs that AMD processors were immune from this specific issue due to the different ways that AMD and Intel processors store and access data:

We are aware of the report of a new security exploit called SPOILER which can gain access to partial address information during load operations. We believe that our products are not susceptible to this issue because of our unique processor architecture. The SPOILER exploit can gain access to partial address information above address bit 11 during load operations. We believe that our products are not susceptible to this issue because AMD processors do not use partial address matches above address bit 11 when resolving load conflicts.

amd-epyc.jpg

SPOILER, one of the latest in the line of speculative execution vulnerabilities that have called into question years of processor architecture design, describes a process that can expose the mappings between virtual and physical memory. That's not a complete issue in and of itself, but it allows other attacks such as Rowhammer to be executed much more quickly and easily.

The research paper that initially disclosed SPOILER earlier this month states that Intel CPUs dating as far back as the first generation Core-series processors are affected. Intel, however, has stated that the vulnerabilities described in the paper can be avoided. The company provided a statement to PC Perspective following our initial SPOILER reporting:

Intel received notice of this research, and we expect that software can be protected against such issues by employing side channel safe software development practices. This includes avoiding control flows that are dependent on the data of interest. We likewise expect that DRAM modules mitigated against Rowhammer style attacks remain protected. Protecting our customers and their data continues to be a critical priority for us and we appreciate the efforts of the security community for their ongoing research.

Source: AMD

AMD's Radeon Software Adrenalin 2019 Edition 19.2.3 mobilizes

Subject: Graphics Cards, Processors | February 25, 2019 - 07:19 PM |
Tagged: Adrenalin Edition, adrenaline 19.2.3, amd, ryzen, Vega

AMD's regular driver updates have a new trick up their sleeves, they now include drivers for AMD Ryzen APUs with a Vega GPU inside.  Today's 19.2.3 launch is the first to be able to do so, and you can expect future releases to as well.  This is a handy integration for AMD users, even if you have a GPU installed you can be sure that your APU drivers are also up to date in case you need them.  For many users this may mean your Hybrid APU + GPU combination will offer better performance than you have seen recently, with no extra effort required from you.

Slide1.JPG

Along with the support for Ryzen APUs you will also see these changes.

Support For:

  • AMD Ryzen Mobile Processors with Radeon Vega Graphics Up to 10% average performance gains with AMD Radeon Software Adrenalin 2019 Edition 19.2.3 vs. 17.40 launch drivers for AMD Ryzen Mobile Processors with Radeon Vega Graphics.
  • Up to 17% average performance gains in eSports titles with AMD Radeon Software Adrenalin 2019 Edition 19.2.3 vs. 17.40 launch drivers for AMD Ryzen Mobile Processors with Radeon Vega Graphics.
  • Dirt Rally 2 - Up to 3% performance gains with AMD Radeon Software Adrenalin 2019 Edition 19.2.3, on a Radeon RX Vega 64 in Dirt Rally 2.

Fixed Issues:

  • Battlefield V players may experience character outlines stuck on screen after being revived.
  • Fan speeds may remain elevated for longer periods than expected when using Tuning Control Auto Overclock or manual fan curve in Radeon WattMan on AMD Radeon VII.
  • ReLive wireless VR may experience an application crash or hang during extended periods of play.
  • Zero RPM will correctly disable in Radeon WattMan on available system configurations when manual fan curve is enabled.
  • A loss of video may be intermittently experienced when launching a fullscreen player application with Radeon FreeSync enabled.

Known Issues:

  • Mouse lag or system slowdown is observed for extended periods of time with two or more displays connected and one display switched off.
  • Changes made in Radeon WattMan settings via Radeon Overlay may sometimes not save or take effect once Radeon Overlay is closed.
  • Some Mobile or Hybrid Graphics system configurations may intermittently experience green flicker when moving the mouse over YouTube videos in Chrome web browser.
    • A work around if this occurs is to disable hardware acceleration.
  • Radeon WattMan settings changes may intermittently not apply on AMD Radeon VII.
  • Performance metrics overlay and Radeon WattMan gauges may experience inaccurate fluctuating readings on AMD Radeon VII.

 

Source: AMD

Report: Intel Pentium Gold G5620 on the Way: First 4 Ghz Pentium

Subject: Processors | February 20, 2019 - 09:59 AM |
Tagged: rumor, report, processor, pentium, Intel, G5620, G5600T, G5420T, G5420, G4950, G4930T, G4930, cpu, celeron

The Pentium processor has been around since the end of the 486 era, introduced in 1993 at a startling cost of $878 for the 60 MHz version, and $964 for 66 MHz (when purchased in quantities of 1000, that is). Now Intel is taking Pentium into uncharted waters for 2019, with the Pentium Gold G5620 reaching 4.0 GHz for the first time for a processor bearing the iconic brand.

Pentium_Gold.jpg

Image via Tom's Hardware

According to reports from Tom's Hardware and AnandTech the Pentium G5620, listed early by retailers in Europe, is a 2-core / 4-thread part that will apparently be at the top of the new budget desktop CPU lineup. Alongside the Pentium G5620 there will refreshed Pentium and Celeron CPUs, as listed by Tom's Hardware:

"...the other processors listed include the G5420 (3.8 GHz, 2/4), G5600T (3.3 GHz, 2/4), G5420T (3.2 GHz, 2/4), the Celeron G4950 (3.3 GHz, 2/2), the Celeron G4930 (3.2 GHz, 2/2), and the Celeron G4930T (3.0 GHz, 2/2)."

We do not have an Intel announcement yet of course, so no details about architecture, process tech, or official pricing. March or April is the expected timeframe based on the listings, and with no official release dates we can only speculate on actual availability here in the U.S.

Rumor: AMD Launching 7nm Ryzen 3000, X500 Motherboards, and Navi GPUs on July 7th

Subject: Processors | February 19, 2019 - 02:57 PM |
Tagged: Zen 2, x570, X500, Ryzen 3000, navi, matisse, amd, 7nm

Spotted by HardOCP, Paul from Red Gaming Tech recently shared leaked information from a source with a reputation of being reliable (from past leaks about 7nm GPUs) who claims that AMD will be announcing a plethora of products at Computex in June to setup for the launch of Zen 2-based 7nm "Matisse" Ryzen 3000 desktop processors, X500 series chipset-based motherboards, and 7nm Navi-based consumer gaming graphics cards on July 7th (The 7th for 7nm I guess).

AMD_Ryzen_3rd_Gen.jpg

Image via AnandTech

As a refresher, Zen 2 is the next major architectural jump for AMD while also pushing a new smaller process node. AMD has not yet revealed all the details about Zen 2 especially about consumer chips, but the new microarchitecture is said to feature tweaks to the front end that along with clockspeed bumps from the TSMC 7nm process will allow them to realize notable IPC and single threaded performance gains. When talking about EPYC 2 "Rome" server processors (Zen 2 based) AMD hinted at changes to branch prediction and pre-fetching as well as increased cache sizes and larger FPUs (256-bit), for example. The move to 7nm allegedly allows AMD to hit similar power envelopes to Zen+ (12nm) Ryzen 2000 series processors while hitting much higher clockspeeds at up to 5.1 GHz boost on their top-end chip. While mobile chips may strike a finer balance between power usage and performance with the move to 7nm, on the desktop AMD is spending nearly all the power savings on performance (which makes sense). Note that it is still not officially official that AMD is using a scaled down EPYC setup with more than one 7nm (TSMC) CPU die and a separate IO die (14nm Global Foundries), [they only teased a chip at CES with an IO die and a single CPU die] but I am of the opinion that that particular rumor makes more sense than otherwise so am inclined to believe this is the case.

AMD Ryzen 3000 7nm CPU and 14nm IO die.png

Ryzen 3000 series processors feature an IO chiplet along with what is rumored to be up to two CPU chiplets (image credit: Tom's Hardware).

From previous leaks, Ryzen 3000 is said to cover all the bases from six core Ryzen 3 3300 series chips to midrange eight core Ryzen 5 and on up to 12 and 16 core Ryzen 9 CPUs that move beyond a single CPU die to two 7nm CPU dies that feature eight cores each. In fact, the top end Ryzen 9 3850X is supposedly a 16 core (32 thread) monster of a desktop chip that has a base frequency of 4.3 GHz and can boost up to 5.1 GHz with a 135W TDP (which when overclocked will likely draw dramatically more like we've seen with both AMD and Intel's top end consumer chips) and price tag of around $520 (400 pounds). The Ryzen 7 3700 and 3700X are 12 core (24 thread) models with TDPs of 95W and 105W respectively with the non-x SKU clocked at 3.8 to 4.6 GHz and the 3700X clocked at 4.2 GHz base and 5 GHz boost. The Ryzen 5 3600 and 3600X are the top end single CPU die models (though a 2x single CCX per die chips might be a reality depending on yields) at eight cores and 16 threads. The Ryzen 3 3300 series parts represent the low end which is now interestingly six cores (oh how times have changed!). Perhaps most interesting of the leaked chips are the Ryzen 5 3600G (~$207) and the Ryzen 3 3300G (~$130) though which feature Navi 12 integrated graphics (presumably these processors combine one 7nm CPU die, one 7nm GPU die, and one 14nm IO die) with 15 and 20 CUs respectively.

As for motherboards, in general the new chips will use the AM4 socket and will be compatible with older 300 and 400 series motherboards with a BIOS update though the top end chips may well necessitate a new X570 or other X500 series motherboard with better power delivery especially for enthusiasts planning to attempt stable overclocks.

Unfortunately, on Navi details are still a bit scarce but the new architecture should bring performance enhancements even beyond Radeon VII (Vega on 7nm). Allegedly due to issues with TSMC, Red Gaming Tech's source believes that Navi might be delayed or pushed back beyond the planned mid-summer release date, but we will have to wait and see. As TSMC ramps up its partial EUV enhanced 7nm node it may free up needed production line space of the current 7nm node for AMD (to fight with others over heh) to meet its intended deadline but we will just have to wait and see!

Take these rumors with a grain of salt as usual but it certainly sounds like it is hoing to be an exciting summer for PC hardware! Hopefully more details about Ryzen 3000 and Navi emerge before then though as that's quite a while yet to wait. Of course, Zen 2 APUs are not coming until at least next year and AMD is still not talking Zen 2 Threadripper which may not see release until the fall at the very earliest. I am very interested to see how AMDs chiplet based design fares and how well they are able to scale it across their product stack(s) as well as what Intel's response will be as it presses on with a fine tuned 14nm++ and a less ambitious 10nm node.

Related reading:

Intel's 28-Core Xeon W-3175X Processor Listed on Newegg

Subject: Processors | February 1, 2019 - 04:37 PM |
Tagged: xeon, workstation, W-3175X, system integrator, SI, processor, parts, OEM, newegg, Intel, DIY, cpu

In a move that would seem to contradict what we have heard about Intel's new 28-core Xeon W-3175X processor, Newegg currently has it listed as a standalone CPU part for $2977.99.

The official announcement from Intel had only mentioned availability via pre-built workstations from system integrators:

"How You Get It: The Intel Xeon W-3175X processor is available from system integrators that develop purpose-built desktop workstations."

newegg_w3175x.png

Product page at Newegg.com

Though not available for purchase (yet?), the existence of this product entry in Newegg's system suggests that the DIY community will have access to Intel's most powerful workstation processor after all, and without a markup over the tray price.

Source: Newegg

Intel's Xeon W-3175X is a 28-Core Workstation Monster

Subject: Processors | January 30, 2019 - 08:13 PM |
Tagged: xeon, workstation, W-3175X, processor, Intel, cpu

Officially unveiled back in October, Intel's newly-launched Xeon W-3175X processor is now available from system integrators, and it is far more extreme than the Xeon name might indicate.

intel-xeon-w-3175x-3.jpg

The Xeon W-3175X in action (image via Intel)

Here is a look at the specs from Intel:

  • Base Clock Speed: 3.1 GHz
  • Maximum Single Core Turbo Frequency: 4.3 GHz
  • Cores/Threads: 28/56
  • TDP: 255W
  • Intel Smart Cache: 38.5 MB
  • Unlocked: Yes
  • Platform PCIE Lanes: Up to 68
  • Memory Support: Six Channels, DDR4-2666
  • Standard RAS Support: Yes
  • ECC Support: Yes
  • RCP Pricing (USD 1K): $2,999

This unlocked 28-core/56-thread CPU offers a base clock speed of 3.1 GHz and Turbo of up to 4.3 GHz (single-thread), and that level of performance comes with a 255-watt TDP. In fact a special cooler from Asetek was also announced today which was developed with Intel for this CPU.

ASETEK_34591.jpg

Image credit: Asetek

And while a $3000 price tag is obviously not going to drive this into mainstream adoption, this processor only being offered through system integrators at this time, and is aimed at the high-end workstation segment. As to performance, there are some day-one reviews out there from GamersNexus, AnandTech, and PC World, among others, and the consensus seems to be that this is an impressive performer, with particular workload the key to performance relative to competing options such as AMD's Threadripper 2990WX (which currently sells for $1730).

xeon_w_3175_pcworld.jpg

Image credit: PC World

We don't have the answers yet about about total platform costs with motherboard pricing currently an unknown, and (more importantly) system integrators the only way to obtain it, but performance in Adobe CS applications alone will likely make this attractive to content creators at the very least.

Source: Intel

AMD Launches 2nd Gen 15W and 35W Ryzen 3000 Mobile Processors with Vega Graphics

Subject: Processors | January 6, 2019 - 03:07 PM |
Tagged: Zen+, vega APU, ryzen mobile, ces2019, athlon, amd

AMD today officially announced its lineup of 2nd generation Ryzen mobile processors, designated Ryzen 3000 Series Mobile Processors. Unlike AMD’s expected 3000-series desktop launch, which will be based on Zen 2, these new mobile variants stick with AMD’s 12nm Zen+ architecture.

Each 15- or 35-watt model features Vega graphics and core/thread counts ranging from 2 cores/4 threads to 4 cores/8 threads. AMD is touting improvements in battery life and overall performance, claiming that the top-end 15-watt part can best the Intel i7-8550U by up to 29 percent in media editing, while the mid-tier 15-watt Ryzen 5 3500U beats its Intel counterpart, the i5-8250U, by up to 14 percent in website loading speed.

Model Cores/Threads TDP Boost/Base Freq. Graphics GPU Cores Max GPU Freq.
AMD Ryzen 7 3750H 4/8 35W 4.0/2.3GHz Vega 10 1400MHz
AMD Ryzen 7 3700U 4/8 15W 4.0/2.3GHz Vega 10 14000MHz
AMD Ryzen 5 3550H 4/8 35W 3.7/2.1GHz Vega 8 1200MHz
AMD Ryzen 5 3500U 4/8 15W 3.7/2.1GHz Vega 8 1200MHz
AMD Ryzen 3 3300U 4/4 15W 3.5/2.1GHz Vega 5 1200MHz
AMD Ryzen 3 3200U 2/4 15W 3.5/2.6GHz Vega 3 1200MHz
AMD Athlon 300U 2/4 15W 3.3/2.4GHz Vega 3 1000MHz
AMD A6-9220C 2/2 6W 2.7/1.8GHz R5 3 cores
192 shaders
720MHz
AMD A4-9120C 2/2 6W 2.4/1.6GHz R4 3 cores
192 shaders
600MHz

The initial batch of laptops featuring Ryzen 3000 Series Mobile Processors will be available in the first quarter from partners Acer, ASUS, Dell, HP, Huawei, Lenovo, and Samsung, with additional product launches coming later in the year.

In addition to its flagship Ryzen mobile lineup, AMD is launching a Zen-based Athlon mobile processor, the Athlon 300U, to target entry-level price points. The company has also announced two new 6-watt A-Series chips aimed at the Chromebook market.

Finally, on the software side, AMD announced that starting this quarter, it will provide Radeon Adrenalin driver support to any laptop with a Ryzen processor and integrated Radeon graphics. This will simplify the driver situation for both consumers and manufacturers, as well as give AMD the ability to directly update gamers’ devices for the latest features and game optimizations.

With AMD getting its arguably less-exciting mobile announcements “out of the way” to start CES, this paves the way for the company to make its big desktop-focused announcements during Dr. Lisa Su’s CES keynote on Wednesday.

Source: AMD

ARM Introduces Cortex-A65AE with SMT for Self-Driving Vehicles and Industrial Automation

Subject: Processors | December 31, 2018 - 09:46 PM |
Tagged: SMT, self driving car, cortex A65AE, armv8-a, arm safety ready, arm

Over the holidays I noticed that ARM released information on a new core design aimed at autonomous driving systems. The Cortex-A65AE is part of the company's Automotive Enhanced lineup and follows on the Cortex-A76AE) with its split-lock and other features that are part of ARM's Safety Ready program.

ARM524_Auto chip diagram_Image_ST2-01.png

Aimed at processors that will be used in self driving cars, advanced driver assistance systems (ADAS), aviation, and industrial automation, the Cortex-A65AE core design integrates several safety and redundancy features that meet ASIL D specifications which is a hazard and risk assessment test for an ISO standard (26262) focused on road vehicle safety. Processors will be able to have up to eight cores and will support SMT with each physical core able to run two threads (at different exception levels and/or under different OSes). The cores can be run independently for performance or in lock step for redundancy and integrity checking comparing each other's calculation results (Split-Lock and Dual Core Lock Step respectively). Using the simultaneous multithreading, two threads on a physical core and operate in lock step mode with two other threads on a different physical shadow core according to Anandtech.

ARM has not yet released full details about the Cortex-A65AE core but it utilizes a 6A65AE4-bit out of order execution pipeline with the. ARMv8-A. It can be customized to suit the needs of ARM's partners so exact chip specifications will differ, but in general Cortex-A65AE cores can have 16 to 64 KB L1 instruction and data caches, 64 to 256 KB L2 cache, an optional L3 cache up to 4MB. Other features include support for ARM TrustZone, ECC memory, and ACP connections for accelerators. The new cores are built with ARM's DynamIQ technology and are slated to be used in chips built on the 7nm process node.

According to ARM, Cortex-A65AE cores are 70% faster in integer performance per core and offer up to 3.5 times the memory throughput and six times the read bandwidth for ACP accelerators versus the existing Cortex-A53 cores. The notable performance jump is likely the result of a combination of moving to a smaller process node, the addition of SMT, and architectural improvements and cache and inter-chip routing optimizations.

ARM is positioning the Cortex-A65AE as complementary to the Cortex-A76AE which is to say that the new core is not a direct replacement for it. While the Cortex-A76AE is high performance, the A65AE is high throughput and both cores reportedly have their place in future ADAS and self-driving cars. The Cortex-A65AE cores can be clustered together to do the initial processing and sensor fusion calculations from all of the inputs from cameras, radar, lidar, and other hardware. From there, clusters including Cortex A76AE chips (or a mix of the two) along with other accelerators can be responsible for making the decisions based on the sensor information. How well it works in practice and how this heterogenous setup will compare to competing offerings from NVIDIA, Intel/MobileEye, and others remains to be seen. I am all for the self-driving car future though so the more competition and developments in that space is always nice to see even if it's still a ways off yet!

The Cortex-A65AE being the first Cortex-A core to feature multithreading is also interesting and I am very curious if we will see that capability expanded to other ARM processors outside of the AE series. While SMT may not be worth it for mobile devices like smartphones and even tablets, perhaps future ARM-powered Always Connecred Windows notebook PCs will use processors with SMT capable cores as it would be easier to justify the extra cost in power and size to include multithreading.

What are your thoughts?

(PS I hope everyone had a safe holiday or at least a good week if you don't celebrate! I am looking forward to 2019 and continuing to serve you with bad puns and allegedlys technology coverage!)

Source: ARM

MSI's holiday miracle, you can now overclock the Athlon 200GE

Subject: Processors | December 26, 2018 - 11:52 AM |
Tagged: overclock, 200GE, amd, msi, b350, b450, AM4

If you happen to have an MSI B450 or B350 motherboard, get out there and grab the latest UEFI BIOS which updates support for AGESA version 1.0.0.6 as it may be pulled soon.  The reason it may not last is because it will let you overclock your Athlon 200GE processor, something which is generally impossible to pull off.  TechSpot tried it out successfully on a variety of MSI boards, such as the Gaming Pro Carbon AC and managed to bump the $55 processor from 3.2GHz to 3.8GHz.  You won't see a huge increase in performance, though you will see some and it makes for an interesting experiment.

2018-12-02-image-2.jpg

"In an unexpected turn of events, it's now possible to overclock the otherwise-locked $55 Athlon 200GE processor. In what appears to be a slip up by MSI, the component maker has enabled Athlon overclocking with their latest BIOS release across its entire AM4 motherboard lineup."

Here are some more Processor articles from around the web:

Processors

Source: TechSpot