Ryzen Locking on Certain FMA3 Workloads

Subject: Processors | March 15, 2017 - 05:51 PM |
Tagged: ryzen, Infinity Fabric, hwbot, FMA3, Control Fabric, bug, amd, AM4

Last week a thread was started at the HWBOT forum and discussed a certain workload that resulted in a hard lock every time it was run.  This was tested with a variety of motherboards and Ryzen processors from the 1700 to the 1800X.  In no circumstance at default power and clock settings did the processor not lock from the samples that they have worked on, as well as products that contributors have been able to test themselves.

ryzen.jpg

This is quite reminiscent of the Coppermine based Pentium III 1133 MHz processor from Intel which failed in one specific workload (compiling).  Intel had shipped a limited number of these CPUs at that time, and it was Kyle from HardOCP and Tom from Tom’s Hardware that were the first to show this behavior in a repeatable environment.  Intel stopped shipping these models and had to wait til the Tualatin version of the Pentium III to be released to achieve that speed (and above) and be stable in all workloads.

The interesting thing about this FMA3 finding is that it is seen to not be present in some overclocked Ryzen chips.  To me this indicates that it could be a power delivery issue with the chip.  A particular workload that heavily leans upon the FPU could require more power than the chip’s Control Fabric can deliver, therefore causing a hard lock.  Several tested overclocked chips with much more power being pushed to them seems as though enough power is being applied to the specific area of the chip to allow the operation to be completed successfully.

This particular fact implies to me that AMD does not necessarily have a bug such as what Intel had with the infamous F-Div issue with the original Pentium, or AMD’s issue with the B2 stepping of Phenom.  AMD has a very complex voltage control system that is controlled by the Control Fabric portion of the Infinity Fabric.  With a potential firmware or microcode update this could be a fixable problem.  If this is the case, then AMD would simply increase power being supplied to the FPU/SIMD/SSE portion of the Ryzen cores.  This may come at a cost through lower burst speeds to keep TDP within their stated envelope.

A source at AMD has confirmed this issue and that a fix will be provided via motherboard firmware update.  More than likely this comes in the form of an updated AGESA protocol.

Source: HWBOT Forums

Gigabyte To Host X99 Champion Challenge on HWBOT

Subject: General Tech | April 22, 2015 - 10:26 AM |
Tagged: overclocking, hwbot, gigabyte, contest

Gigabyte will host the upcoming X99 Champion Challenge beginning May 1st on HWBOT.org, and the overclocking contest runs in six stages ending on May 31.

image002.jpg

According to Gigabyte, "by participating, overclockers have the chance to win $2,800 USD in cash prizes and some exciting hardware, including the leader of them all, the X99-SOC Champion!" True to the name of the contest participants must use a Gigabyte X99 motherboard, and each stage offers a different challenge:

Contest Stages
Stage 1: XTU - May 1st until May 8th, 2015
- CPU frequency 4GHz max
- RAM at 3300MHz max.
- GIGABYTE X99 Motherboards only

Stage 2: XTU - May 8th until May 15th, 2015.
- CPU frequency 4.5GHz max
- Uncore at 4.5GHz max.
- GIGABYTE X99 Motherboards only

Stage 3: XTU - May 15th until May 31st, 2015.
- CPU frequency 5GHz max
- GIGABYTE X99 Motherboards only

Stage 4: Fire-Strike - May 1st until May 27th, 2015.
- Single NVIDIA GT 730 graphics card
- GIGABYTE X99 Motherboards only

Stage 5: Catzilla 720p - May 1st until May 28th, 2015.
- Single NVIDIA GT 730 graphics card
- GIGABYTE X99 Motherboards only

Stage 6: 3DMark 2001 SE - May 1st until May 28th, 2015.
- Single NVIDIA GT 730 graphics card
- GIGABYTE X99 Motherboards only

The full press release with contest rules is available here.

Source: Gigabyte

Windows 8 RTC Bug Found, Benchmark Results Still Banned At HWBot

Subject: General Tech | August 27, 2013 - 07:59 PM |
Tagged: windows 8, overclocking, hwbot

Earlier this month, competitive overclocking website HWBot banned benchmark results that come from systems that use Microsoft’s latest Windows 8 operating system. Unfortunately, Microsoft made a compromise and choose to use an internal software timer rather than a hardware-based Real Time Clock (RTC) to ensure compatibility with budget and embedded systems that skip such hardware to cut costs. Ocaholic’s Christian Ney has further analyzed the RTC bug and posted an article on the bug and how to fix it.

HWBot’s issue with Windows 8 is that users are able to artificially inflate benchmark scores by down-clocking the CPU BCLK using software in Windows 8 rather than via the BIOS (which does not effect results). Under Windows 8, when down-clocking using software tools, the amount of time the benchmark measures does not match up with the amount of time that has actually passed--and would be correctly reported by a hardware timer. Unfortunately, Windows 8’s Real Time Clock references the QueryPerformanceCounter (QPC) which is able to read a hardware timer but does not by default on Intel-based platforms. On the other side of things, AMD-based systems are less effected by this bug, but users could turn off the default hardware timer to cause similar artificial benchmark score inflation (the system is able to push out more calculations and/or frames per second in a reported amount of time that is less than the actual time it took to complete the benchmark, giving users an artificially higher score).

For Intel users, the bug can reportedly be fixed by opening up an administrator command prompt and typing the following command.

“bcdedit /set {current} useplatformclock Yes”

While it is nice to see the analysis and a fix for the bug, HWBot is still not allowing Windows 8 benchmark scores. An update on HWBot's original blog post states that all new Windows 8 results except those from non-overclockable notebooks are banned. Further, any existing benchmarks from Windows 8 systems that seem out of line or are either world records or top scores will be removed from the site. Windows 8 benchmarks with AMD hardware are also banned.

It appears that, although there is a fix for the RTC bug, until HWBot can somehow verify that users have either implemented the fix on Intel systems or not changed the default settings for AMD systems, benchmark results from Windows 8 can still not be trusted. Here’s hoping some method will become available to allow for that necessary verification.

Source: HWBot