Language Selection

English French German Italian Portuguese Spanish

Graphics/Benchmarks

AMD Ryzen Threadripper 3970X / 3960X Linux Benchmarks

Filed under
Graphics/Benchmarks

After the embargo on the Intel Core i9 10980XE expired a few hours ago, now we are allowed to share the performance numbers on the new AMD Ryzen Threadripper 3960X and 3970X processors. These new Zen 2 HEDT CPUs pack a real performance punch, but do come in as more expensive than the i9-10980XE and there is one boot-stopping Linux bug to mention with a workaround... But besides that lone Linux support caveat, the Threadripper 3960X and Threadripper 3970X absolutely dominate in performance.

The Ryzen Threadripper 3960X is AMD's new $1399 USD processor that features 24 cores / 48 threads, 3.8GHz base frequency, and 4.5GHz boost frequency with an impressive 140MB cache. This isn't even the top-end Zen 2 HEDT and in core count and most other details already outpaces the Core i9 10980XE: the base clock is 800MHz higher than the 10980XE but the turbo/boost clock is 100MHz lower on the 3960X.

Read more

Intel Core i9 10980XE Linux Performance Benchmarks

Filed under
Graphics/Benchmarks

Intel today is rolling out the Core i9 10980XE as their new Cascade Lake X-Series processor that features 18 cores / 36 threads with a maximum turbo frequency of 4.6GHz and TBM 3.0 frequency of 4.8GHz. Following a last minute change, Intel moved up the embargo lift time of the Core i9 1980XE so here are the results we can share with you right now.

The Intel Core i9 10980XE Cascade Lake processor features the same core / thread count as the previous Core i9 9980XE and i9 7980XE but now with a 3.0GHz base frequency, 4.6GHz peak turbo frequency, 4.8GHz Turbo Boost Max 3.0 frequency, DDR4-2933 quad channel memory support rather than DDR4-2666, and the L1TF/Meltdown hardware mitigations in place. The cache size remains the same at 24.75MB and the processor having a 165 Watt TDP.

Read more

Linux Graphics: NVIDIA and AMD

Filed under
Graphics/Benchmarks
Linux
  • The Open-Source NVIDIA Driver Will End Out 2019 In Poor Shape Still For Newer GPUs

    For the imminent Linux 5.5 kernel cycle we have talked about exciting AMD Radeon and Intel graphics driver changes on deck from Navi OverDrive overclocking to more Intel Tiger Lake and Jasper Lake bits, AMDGPU HDCP support, and other features queued. But what about the open-source NVIDIA "Nouveau" Linux driver?

    Sadly, it's been several kernel cycles since there has been anything major to report. In fact, as it stands right now, there haven't been any changes queues from Nouveau into DRM-Next. Earlier this week Red Hat's Ben Skeggs did push a few bits of new code to the skeggsb/nouveau repository but that work hasn't appeared in DRM-Next as of writing nor any indication on the mailing list.

  • AMDKFD/ROCm GPU Compute Can Work On POWER Systems Like Raptor's Talos II

    While NVIDIA graphics in IBM POWER systems have been known to make a powerful combination for supercomputer deployments, for those wanting a libre GPU compute experience can also use POWER with AMD Radeon's open-source driver with a pending patch to the kernel driver.

    With various Radeon driver bugs in the open-source stack having been worked out over time that affect the POWER architecture, it turns out the driver stack is good enough on POWER to even enable the AMDKFD (Kernel Fusion Driver) compute support -- which is the kernel component to the Radeon Open Compute (ROCm) stack that runs in user-space.

The Combined Impact Of Mitigations On Cascade Lake Following Recent JCC Erratum + TAA

Filed under
Graphics/Benchmarks

Following the initial tests earlier this month from the disclosures of the JCC Erratum (Jump Conditional Code) that required updated Intel CPU microcode to address and on the same day the TSX Async Abort (TAA) vulnerability that required kernel mitigations to address, which I have run benchmarks of those CPU performance impacts individually, readers have requested tests looking at the current overall impact to the mitigations to date.

In this article are benchmarks for Intel Cascade Lake using dual Xeon Platinum 8280 processors. Keep in mind Cascade Lake has hardware mitigations in place for L1TF, Meltdown, and some Spectre protections. The different configurations tested for this article included...

Read more

Hardware: Benchmarks, 'Internet of Things' and Clientron

Filed under
Graphics/Benchmarks
Hardware
  • POWER9 Blackbird Performance On Ubuntu 19.04 vs. Ubuntu 19.10 Benchmarks

    We have done a lot of benchmarks on Intel/AMD x86_64 for Ubuntu 19.10 for seeing how its performance is looking, but what about IBM POWER9 with the likes of the libre Raptor Blackbird? Here are some Ubuntu 19.04 vs. 19.10 POWER benchmarks I recently carried out. 

    Using the great Blackbird POWER9 system from Raptor Computing Systems, I compared the performance of clean / out-of-the-box installs of Ubuntu 19.04 and Ubuntu 19.10. 

  • How many IoT devices do you own?

    In the age of the Internet of Things, how connected are the devices in your home or office? Take our poll.

  • All-in-one 23.8-inch thin client integrates SDM-L computer

    Clientron’s “TC-238 AIO Thin Client” has a 23.8-inch display and an Atom x5-E38000 SoC implemented via Intel’s SDM-Large form factor. Triple simultaneous displays are available thanks to DVI-D and 4K-ready DP ports.

    Thin client vendor Clientron, which last year released an S-Cube Pi 3 B+ Thin Client built around the Raspberry Pi 3 B+, has jumped on another tech trend by announcing a fanless TC-238 AIO Thin Client that adopts Intel’s Smart Display Module-Large (SDM-L) form factor.

Graphics: Mesa, Linux 5.6 and X.Org Server 1.20.6

Filed under
Graphics/Benchmarks
  • Various Game Emulators Are Faster On Mesa Drivers Now Thanks To OpenGL Threading

    A few days ago 7 Days to Die saw a performance boost on Mesa Git from its "glthread" threading implementation while now a number of game emulators have seen similar whitelisting.

    The Dolphin GameCube/Wii emulator sees about 17% better performance from Mesa OpenGL threading, the Citra Nintendo 3DS emulator saw a 12% rise in performance, and the Yuzu Nintendo Switch emulator saw a 29% rise in performance from this whitelisting.

  • Linux 5.6 Will Bring Another Radeon Run-Time Power Management Improvement

    While the Linux 5.5 cycle begins next week, looking ahead to early next year when the Linux 5.6 cycle will begin, there is expected to be another power management improvement coming for AMD Radeon graphics cards.

    Right now the Linux kernel doesn't support run-time power management for AMD display audio hardware meaning wasted energy when not using the DisplayPort / HDMI audio. As part of the recent work on the AMDGPU driver's "BACO" (Bus Active, Chip Off) support, run-time power management for the audio hardware has been in the works.

  • X.Org Server 1.20.6 Released With Many Bug Fixes - Helps XWayland, PRIME + Other Bits

    X.Org Server 1.20.6 has some fixes around PRIME GPU handling, XWayland now expands the RandR screen size limits, modesetting DDX driver fixes, various XWayland fixes, GLX vendor selection support courtesy of NVIDIA, syncing of the Intel PCI IDs from Mesa for the DRI code, and a variety of other fixes. There are ten bug fixes to XWayland code in 1.20.6 making for this being a fairly notable update.

  • xorg-server 1.20.6
    A variety of bugfixes across the board, primarily in Xwayland and PRIME.
    This release also includes GLX vendor selection support. Thanks to all
    who contributed with testing and fixes!
    
    Aaron Plattner (3):
          GLX: Set GlxServerExports::{major,minor}Version
          xfree86: Call ScreenInit for protocol screens before GPU screens
          os: Don't crash in AttendClient if the client is gone
    
    Adam Jackson (7):
          xwayland: Expand the RANDR screen size limits
          miext/sync: Fix needless ABI change
          glx: Fix previous context validation in xorgGlxMakeCurrent
          meson: Fix another reference to "gl" 9.2.0
          meson: Apparently 1.2 is < 1.2.0
          mi: Add a default no-op miSourceValidate
          dix: Call SourceValidate before GetImage
    
    Alex Goins (5):
          xsync: Add resource inside of SyncCreate, export SyncCreate
          randr: Fix RRCrtcDetachScanoutPixmap() segfault during server teardown
          modesetting: Fix ms_covering_crtc() segfault with non-modesetting slave primary
          modesetting: Fix ms_covering_crtc() segfault with non-xf86Crtc slave
          modesetting: Implement ms_covering_randr_crtc() for ms_present_get_crtc()
    
    Alexander Tsoy (1):
          configure: Set libdrm flags correctly if only XORG is enabled
    
    Alexander Volkov (1):
          shm: Use memfd_create when possible
    
    Andres Rodriguez (1):
          xf86: Disable unused crtc functions when a lease is revoked
    
    Carlos Garnacho (4):
          xwayland: Reset scheduled frames after hiding tablet cursor
          xwayland: Separate DamagePtr into separate window data
          xwayland: Refactor surface creation into a separate function
          xwayland: Handle the case of windows being realized before redirection
    
    Eric Anholt (2):
          shm: reindent shm_tmpfile to follow our standards.
          shm: Pick the shm dir at run time, not build time.
    
    Hans de Goede (1):
          glamor/xwayland: Define EGL_NO_X11
    
    Kyle Brenneman (3):
          GLX: Add a per-client vendor mapping.
          GLX: Use the sending client for looking up XID's
          GLX: Add a function to change a clients vendor list.
    
    Marco Trevisan (Treviño) (1):
          Xi: Use current device active grab to deliver touch events if any
    
    Marvin Schmidt (1):
          build: glx: Lower gl version to work with libglvnd
    
    Matt Roper (1):
          dri2: Sync i965_pci_ids.h from mesa
    
    Matt Turner (3):
          dix: Assert noPanoramiXExtension is false in PanoramiX code
          xfree86: Test presence of isastream()
          xserver 1.20.6
    
    Michel Dänzer (2):
          Revert "present/scmd: Check that the flip and screen pixmap pitches match"
          miext/sync: Make struct _SyncObject::initialized fully ABI compatible
    
    Olivier Fourdan (7):
          xwayland: Avoid a crash on pointer enter with a grab
          xwayland: Check status in GBM pixmap creation
          glamor: Make pixmap exportable from `gbm_bo_from_pixmap()`
          xwayland: Update screen pixmap on output resize
          xwayland: Do not free a NULL GBM bo
          compiler.h: Do not include sys/io.h on ARM with glibc
          present/wnmd: Relax assertion on CRTC on abort_vblank()
    
    Samuel Thibault (2):
          Fix crash on XkbSetMap
          Fix crash on XkbSetMap
    
    Sven Joachim (1):
          modesetting: Fix broken manpage in autoconf build
    
    git tag: xorg-server-1.20.6
    

RADV's ACO Back-End Is Helping Radeon Navi Linux Gaming Performance

Filed under
Graphics/Benchmarks
Linux
Gaming

It's been almost two months since last looking at the RADV ACO performance for this shader compiler back-end alternative to the AMDGPU LLVM code. ACO is making its debut in the upcoming Mesa 19.3 release while since the last round of testing have been more optimizations and fixes as well as getting the Navi/GFX10 support in place. In this article are some fresh benchmarks of the Vulkan RADV ACO support for not only Polaris and Vega but also the Radeon RX 5700 Navi graphics cards.

With a Radeon RX 590, RX 5700, RX 5700 XT, and VII graphics cards I tested Mesa 20.0-devel Git as of this week paired with Linux 5.4 Git -- the Linux 5.4 kernel upgrade also helps gaming performance. This is our first look at the ACO performance for new Navi graphics processors plus a fresh look in general with the ever-evolving state of Mesa Git. There are also new game tests included, namely the recently ported Shadow of the Tomb Raider by Feral Interactive for seeing how that benefits from the ACO back-end.

Read more

Mesa 19.2.6

Filed under
Graphics/Benchmarks

Hi list,

I'm releasing a new mesa 19.2.x release to address being unable to compile on
PPC due to a bad backport. There are a couple of additional patches in here
because I didn't want to tease them apart and they're all stable anyway.

Dylan

Shortlog
========

Alejandro Piñeiro (1):
      v3d: adds an extra MOV for any sig.ld*

Dave Airlie (1):
      llvmpipe/ppc: fix if/ifdef confusion in backport.

Dylan Baker (4):
      docs/relnotes/19.2.5: Add SHA256 sum
      meson: generate .pc files for gles and gles2 with old glvnd
      docs: Add release notes for 19.2.6
      VERSION: bumpre to 19.2.6

Eric Engestrom (1):
      vulkan: delete typo'd header

Hyunjun Ko (1):
      freedreno/ir3: fix printing output registers of FS.

Jose Maria Casanova Crespo (1):
      v3d: Fix predication with atomic image operations

Yevhenii Kolesnikov (1):
      glsl: Enable textureSize for samplerExternalOES



git tag: mesa-19.2.6

Read more

Also: Mesa 19.2.6 Released Due To POWER Fallout

Nvidia Outs New Linux/BSD Graphics Driver with GeForce GTX 1650 SUPER Support

Filed under
Graphics/Benchmarks
Linux

For Linux- and BSD-based platforms, the Nvidia 440.36 proprietary graphics driver is here to add support for the Nvidia GeForce GTX 1650 SUPER graphics card, which Nvidia claims it's up to 50 percent faster than the original GTX 1650 and up to 2X faster than the previous-generation GTX 1050.

Now BSD and Linux gamers who bought an Nvidia GeForce GTX 1650 SUPER GPU can use it to play games at full performance if they install the Nvidia 440.36 proprietary graphics driver, which is available to download only for 64-bit operating systems from Nvidia.com or via our free software portal here and here.

Read more

A Look At The GCC Compiler Tuning Performance Impact For Intel Ice Lake

Filed under
Graphics/Benchmarks

For those wondering if it's worthwhile for performance recompiling your key Linux binaries with the microarchitecture instruction set extensions and tuning for Ice Lake, here are some GCC compiler benchmarks looking at that impact for the Core i7 1065G7 on the Dell XPS 7390.

In particular, this article is looking at the affect on generated benchmark binaries when built under the following CFLAGS/CXXFLAGS configurations:

-O3 -march=skylake - Just optimizing for conventional Skylake processors.

-O3 -march=skylake-avx512 - Optimizing for Skylake AVX-512 processors like Skylake-SP/Skylake-X. The Skylake AVX-512 enables use of the AVX512F, CLWB, AVX512VL, AVX512BW, AVX512DQ and AVX512CD instructions.

-O3 -march=icelake-client - Optimizing for Icelake client/desktop processors. New instructions exposed here not found with Skylake/Skylake-AVX512 include AVX512VBMI, AVX512IFMA, SHA, CLWB, UMIP, RDPID, GFNI, AVX512VBMI2, AVX512VPOPCNTDQ, AVX512BITALG, AVX512VNNI, VPCLMULQDQ, and VAES. Note there is also the "icelake-server" target for future Ice Lake Xeon Scalable processors where additionally PCONFIG and WBNOINVD are flipped on.

Read more

Syndicate content