Tag Archives: AMD Radeon

AMD resolved PCIe power draw issue on RX480 with new driver

I’ve posted initial findings about the excessive PCIe power draw issue here. AMD promised a fix for it and today they’ve delivered it in a form of a new driver.

Like I’ve predicted, modern graphic cards have very flexible power delivery system and Polaris is no different. What AMD did here was rearrange the power delivery between PCIe slot and 6pin power connector. Now, Radeon RX480 draws power from PCIe slot within specified limits and draws a bit more from 6pin. Officially 6pin is rated at 75W, but can realistically deliver up to 150W. So, redirecting power like this solves the initial problem while doesn’t impair performance at all.

AMD did provide additional “Compatibility Mode” which restricts power usage even further. I frankly don’t think anyone should ever enable this, but if you feel like saving some extra watts, you can enable it in Radeon Settings.

To verify the fix, TechPowerUp ran a test and everything is in order just like AMD promised.

You can read the TechPowerUp re-test here.

The PCIe power draw fix is included in AMD Radeon Crimson Edition version 16.7.1, so make sure you upgrade the drivers asap. Btw, AMD dropped a small 3% performance boost for popular games in this driver so even if you use Compatibility Mode, you shouldn’t see any performance difference.

I really like the way things turned out here. Firstly, for reviewers to point out the issue and secondly for AMD to professionally fix it. An the ones who benefit from all this the most are us, the customers. Yay 🙂

Everyone losing their shit about RX480 power consumption

I’m gonna drop a quick post about this, because people apparently aren’t capable of thinking rational anymore. Also the double standards when it comes to NVIDIA and AMD…

As you might have heard, AMD released Radeon RX480 recently, a killer cheap graphic card based on new Polaris architecture. It’s priced up to $230 for 8GB version and performs a bit better than GTX 970/R9 390, but slightly worse than GTX 980 or R9 390X. Not bad to be honest.

Things got complicated when reviewers found out some cards draw more than 150W of power. Now, that by itself wouldn’t be a problem if the excess power draw wasn’t pulled from PCIe slot which is rated at 75W. The 6pin PCIe power connector on the graphic card is rated at 75W as well officially, but can draw a lot more power without any problem. Both combined deliver 150W of power. But the card in tests pulled 166W. So, you have to get more power from somewhere and RX480 apparently does it from PCIe slot. Pulling more than 75W from PCIe slot can potentially damage cheap crappy motherboards.

So, up till this point, I have no problems. There apparently is an issue and reviewers are there to point it out so company behind the product is aware of the issue and is going to remedy it. After all, I’m a consumer and such behavior is preferred because it benefits consumers.

However (not unexpected eh?), AMD already officially acknowledged the issue with this statement (provided by W1zzard from TechPowerUp):

“As you know, we continuously tune our GPUs in order to maximize their performance within their given power envelopes and the speed of the memory interface, which in this case is an unprecedented 8 Gbps for GDDR5. Recently, we identified select scenarios where the tuning of some RX 480 boards was not optimal. Fortunately, we can adjust the GPU’s tuning via software in order to resolve this issue. We are already testing a driver that implements a fix, and we will provide an update to the community on our progress on Tuesday (July 5, 2016).”

As you can see, AMD is already preparing a fix for it. Modern GPU’s are very advanced and power delivery can be fully controlled via BIOS or driver. I don’t know exact Polaris electrical design, but knowing Maxwell 2 already has this, been fiddling with it myself on GTX 980 that I have, and I see no reason why brand new GPU like Polaris wouldn’t have it as well. Meaning AMD isn’t just selling hot air and buying time, they have a realistic solution. Tuesday 5th July is the day they will give more info and potentially a driver with the fix. They can either strictly restrict power draw to 150W as whole or restrict PCIe to 75W as specified and let 6pin additional connector draw a bit of excess power. From what I heard, even though 6pin power connector is rated to 75W, it can pull up to 150W just like 8pin. Which won’t “gimp” the performance, it will just bring it to level AMD has specified while leaving PCIe within specs.

Has that calmed people? Nope. Everyone still losing their shit and creating more drama even before anyone can even evaluate the fix. I’m pretty sure using graphic card in such conditions for few days won’t affect anything. So why all this fucking drama?

What’s even more hilarious, NVIDIA had the same shit. On TWO occasions and I’ve only heard about it now. Never before. NVIDIA also fucked up the GTX 1080 fan profiles on Founders Edition cards (reference models)? There were mentions of it, but nowhere on the same level of crazy nonsense people are doing now for AMD.

For fucks sake, stop being such god damn fanboys. I own a GTX 980 and I’m defending AMD here…

Everyone calm the fuck down and wait for the fix. Evaluate it and if performance or anything else will be greatly affected by that, then start losing your shit again. But until then, calm the fuck down. Fucking hell.

UPDATE (2016/07/06)!

AMD issued an update on the given matter like an hour ago on Facebook.

We promised an update today (July 5, 2016) following concerns around the Radeon RX 480 drawing excess current from the PCIe bus. Although we are confident that the levels of reported power draws by the Radeon RX 480 do not pose a risk of damage to motherboards or other PC components based on expected usage, we are serious about addressing this topic and allaying outstanding concerns. Towards that end, we assembled a worldwide team this past weekend to investigate and develop… a driver update to improve the power draw. We’re pleased to report that this driver—Radeon Software 16.7.1—is now undergoing final testing and will be released to the public in the next 48 hours.

In this driver we’ve implemented a change to address power distribution on the Radeon RX 480 – this change will lower current drawn from the PCIe bus.
Separately, we’ve also included an option to reduce total power with minimal performance impact. Users will find this as the “compatibility” UI toggle in the Global Settings menu of Radeon Settings. This toggle is “off” by default.

Finally, we’ve implemented a collection of performance improvements for the Polaris architecture that yield performance uplifts in popular game titles of up to 3%. These optimizations are designed to improve the performance of the Radeon RX 480, and should substantially offset the performance impact for users who choose to activate the “compatibility” toggle.

AMD is committed to delivering high quality and high performance products, and we’ll continue to provide users with more control over their product’s performance and efficiency. We appreciate all the feedback so far, and we’ll continue to bring further performance and performance/W optimizations to the Radeon RX 480.

Interestingly enough, they will provide the fix, but they are confident enough the problem with PCIe power draw isn’t serious enough to enable the fix by default. Which is a bit strange, but I guess they know what they are doing. They also optimized drivers for a 3% boost which offsets the roughly 1% performance penalty when enabling the fix. Meaning even if users decide to enable the fix, they won’t lose any performance. I’m still interested in seeing performance and power draw results in a re-test of the RX480 with new drivers compared to old ones (and with or without the fix). Just to be really sure what’s happening. I’ll keep you posted…

UPDATE (2016/07/08)!

Read the news about resolved PCIe power issues on RX480 here. I’ve decided to post it as a new article while linking it back here for reference.

AMD fixed the issue entirely. That’s what I call a professional response for seemingly unfixable problem…

 

AMD Radeon R9-380 starting to reveal itself…

amd_radeon_artwork_angle_new1It is now official. AMD Radeon R9-380 is planned for Q2-2015, roughly around March. It will sport whooping 4096 stream processors based on Tonga design (GCN 1.2) on a single die, new framebuffer compression for increased bandwidth (also from Tonga GPU), all paired to 4GB of HBM stacked VRAM (Video RAM). Nothing said on the GPU clocks, but I hope it will be 1,2 GHz. Memory clocks on HBM memory are lower, clocking at just 1,2GHz, but they offer massive bandwidth of 128GB/s. Doesn’t sound much on its own, but each Radeon R9-380 graphic card will feature 4 units, totaling 512GB/s bandwidth. And that’s before even taking framebuffer compression into account! R9-290 only has 320GB/s, so do the math yourself 😀

Was already looking at GTX 970, but considering my HD7950 is still doing pretty well I thought, why not wait these two more months and have an option to decide between two different GPU’s. If Radeon turns out to be awesome, that’s what I’ll take. If it’ll be “meh”, then I’ll take the GTX 970, depending on the price of course. Either way it’s win-win for consumers 😀

AMD Catalyst Omega 14.12 WHQL released!

So, the highly anticipated AMD Catalyst Omega drivers have finally landed. And yes, they actually have Omega in the name and no, they are not in any way related to the old community modded Omega drivers. They feature several interesting features like…

VSR (Virtual Super Resolution)

AMD_VSR_SLIDE1Basically this is the same as NVIDIA’s DSR. It renders the image at 4K and downscales it to 1080p on the output. It convinces games that your display can render higher resolution than it actually supports, so the game has to offer you option for 4K. All fine and well, except it only works with R9 290 cards. Srsly? Where is that “all GCN cards will get the same treatment” from the past?. HD7900 series are GCN powered. And yet we got screwed. It’s good that in the future I’ll get this feature if I decide to buy another Radeon, but why screw the existing HD7900 users? Which also includes R9-280 users because it is based upon HD7900 series. I’ve been with AMD for the last several generations of GPU’s. Do you think I’d not buy another Radeon if I get VSR now on my HD7950? If that was the case, I’d already have GTX 970 in my box. But I don’t, because I’ve decided to wait for AMD’s response with R9-300 series when it gets out. Maybe HD7900 doesn’t have the 4K grunt for modern games, but there are older games that we play and that would greatly benefit from VSR. Anyway…

Improved Frame Pacing

AMD_FP_SLIDE2Supposedly only for multi card configurations, improving many games smoothness. They list Tomb Raider, Metro series, Batman Arkham Origins, Sniper Elite 3 and more. We’ll see if it makes any difference on single GPU…

AMD Fluid Motion Video

AMD_FLUIDMV_SLIDE3This feature is basically what high end LCD TV’s do, interpolate missing frames and make videos super smooth and sharp during motion. Something like Philips Perfect Natural Motion or LG’s TruMotion. Only downside is that it only works in Cyberlink PowerDVD 14. Which makes it kinda useless considering I never used that thing and never will. Media Player Classic Home Theater (MPC HC) or VLC support would make tons more sense…

Video filtering and 1080p Detail Enhancement

AMD_CR_SLIDE4AMD_1080PENH_SLIDE5These features enhance video quality and are available for regular common media players that use DXVA and also work on HD7900 series. Which is nice.

4K Ultra HD Upscaling

AMD_UHD_SLIDE6It gives ability to upscale 1080p video into 4K, enhance it and display it on your 4K monitor or 4K TV. I have a 4K Philips LCD, but since I’m on HD7900, I’m out of luck yet again…

Performance Enhancements

AMD claims 19% average framerate improvements which sounds like a lot, but TechPowerUp tests proved a bit differently. You can check it out here: Synthetic tests Game tests

While the differences aren’t all that big I’m still going to use it since I use latest drivers all the time. It has worked well so far so why not. I just wish they’d focus less on R9 only. I get it, it’s their current high end, but by supporting your older userbase, you’re showing them your long term commitment. Showing that you care. And it means a lot to me and is a great factor when I’m considering my next upgrade. If I’m being treated well as a customer even when I don’t have the latest and most expensive graphic card, I sure will buy a new high end card from the same company when I’ll consider my next upgrade. So far AMD showed decent support, but I’d wish  for a bit more…

Download:

http://support.amd.com/en-us

AMD Catalyst 14.6 BETA released!

AMD has released a new Catalyst driver package version 14.6 BETA. Most significant changes are removal of the Windows 8 support, so make sure you update your Windows to Windows 8.1 Update 1 which is currently the latest version. It’s just so much better, so I’m wondering why you haven’t done so already 😛 Other important changes are quite big performance improvements, mostly for R9-290 series, but maybe even older HD7xxx users will benefit from it. Who knows, I will update anyway as I always do. Oh and AMD also added support for GPU accelerated JPEG images decoding, resulting in twice as fast JPEG images rendering. Too bad only a limited set of APU’s and GPU’s is supported, so I don’t think my older AMD E-450 will be getting any love from it.

Feature Highlights of The AMD Catalyst 14.6 Beta Driver for Windows

  • Performance improvements
    • Watch Dogs performance improvements
      • AMD Radeon R9 290X – 1920×1080 4x MSAA – improves up to 25%
      • AMD Radeon R9290X – 2560×1600 4x MSAA – improves up to  28%
      • AMD Radeon R9290X CrossFire configuration (3840×2160 Ultra settings, MSAA = 4X)  – 92% scaling
    • Murdered Soul Suspect performance improvements
      • AMD Radeon R9 290X – 2560×1600 4x MSAA – improves up to 16%
      • AMD Radeon R9290X CrossFire configuration (3840×2160 Ultra settings, MSAA = 4X)  – 93% scaling
  • AMD Eyefinity enhancements:
    • Mixed Resolution Support
      • A new architecture providing brand new capabilities
      • Display groups can be created with monitors of different resolution  (including difference sizes and shapes)
      • Users have a choice of how surface is created over the display group
        1. Fill – legacy mode, best for identical monitors
        2. Fit – create the Eyefinity surface using best available rectangular area with attached displays.
        3. Expand – create a virtual Eyefinity surface using desktops as viewports onto the surface.
    • Eyefinity Display Alignment
      • Enables control over alignment between adjacent monitors
    • One-Click Setup
      • Driver detects layout of extended desktops
        1. Can create Eyefinity display group using this layout in one click!
  • New user controls for video color and display settings
    • Greater control over Video Color Management:
      • Controls have been expanded from a single slider for controlling Boost and Hue to per color axis
    • Color depth control for Digital Flat Panels (available on supported HDMI and DP displays)
      • Allows users to select different color depths per resolution and display
  • AMD Mantle enhancements
    • Mantle now supports AMD Mobile products with Enduro technology
      • Battlefield 4:  AMD Radeon HD 8970M (1366×768; high settings) – 21% gain
      • Thief: AMD Radeon HD 8970M (1920×1080; high settings) – 14% gain
      • Star Swarm:  AMD Radeon HD 8970M (1920×1080; medium settings) – 274% gain
    • Enables support for Multi-GPU configurations with Thief (requires the latest Thief update)
  • AMD AM1 JPEG decoding acceleration
    • JPEG decoding acceleration was first enabled on the A10 APU Series in AMD Catalyst 14.1 beta, and has now been extended to the AMD AM1 Platform
    • Provides fast JPEG decompression
    • Provides Power Efficiency for JPEG decompression

Known Issues  

  • Running Watch Dogs with a R9 280X CrossFire configuration may result in the application running in CrossFire software compositing mode
  • Enabling Temporal SMAA in the AMD Catalyst Control center under a CrossFire configuration when playing Watch Dogs will result in Flickering
  • AMD CrossFire configurations with Eyefinity enabled will see application stability with BattleField 4 or Thief when running Mantle
  • Catalyst Install Manager text is covered by Express/Custom radio button text
  • Express Uninstall does not remove C:\Program Files\(AMD or ATI) folder

Download:

http://support.amd.com/en-us/download/desktop?os=Windows%208.1%20-%2064#amd-catalyst-packages

AMD driver webpage is experiencing some weird problems, which makes only a 64bit version available for download. I hope they will soon fix the webpage problems…

AMD Catalyst 14.4 WHQL released!

AMD has finally released the very latest Catalyst 14.4 drivers that are WHQL certified.

Release notes:

http://support.amd.com/en-us/kb-articles/Pages/AMDCatalyst14-4WINReleaseNotes.aspx

Highlights of AMD Catalyst™ 14.4 Windows Driver:

    • Support for the AMD Radeon R9 295X
    • CrossFire fixes enhancements:
      • Crysis 3 – frame pacing improvements
      • Far Cry 3 – 3 and 4 GPU performance improvements at high quality settings, high resolution settings
      • Anno 2070 – Improved CrossFire scaling up to 34%
      • Titanfall – Resolved in game flickering with CrossFire enabled
      • Metro Last Light – Improved Crossfire scaling up to 10%
      • Eyefinity 3×1 (with three 4K panels) no longer cuts off portions of the application
      • Stuttering has been improved in certain applications when selecting mid-Eyefinity resolutions with V-sync Enabled
    • Full support for OpenGL 4.4
      • OpenGL 4.4 supports the following extensions:
        • ARB_buffer_storage
        • ARB_enhanced_layouts
        • ARB_query_buffer_object
        • ARB_clear_texture
        • ARB_texture_mirror_clamp_to_edge
        • ARB_texture_stencil8
        • ARB_vertex_type_10f_11f_11f_rev
        • ARB_multi_bind
        • ARB_bindless_texture
        • ARB_spare_texture
        • ARB_seamless_cubemap_per_texture
        • ARB_indirect_parameters
        • ARB_shader_group_vote
    • Mantle beta driver improvements:
      • BattleField 4: Performance slowdown is no longer seen when performing a task switch/Alt-tab
      • BattleField 4: Fuzzy images when playing in rotated SLS resolution with an A10 Kaveri system

Known Issues:

    • System will TDR or BSOD when encoding with Power Director 11
    • Driver installation might result in a black screen when installing on a Dual AMD Radeon R9 295X configuration under Windows 8.1.   The issue can be overcome by rebooting the PC; upon reboot the display driver will be installed.  The remaining Catalyst components can then be installed.

 

Download:

Win7 / Win8 32bit

Win7 / Win8 64bit

AMD Catalyst 14.4 RC released!

AMD has now officially released the Release Candidate (RC) version of the Catalyst 14.4 drivers that I have posted few days ago.

Feature Highlights of The AMD Catalyst 14.4 Release Candidate Driver for Windows

    • Support for the AMD Radeon™ R9 295X
    • CrossFire™ fixes enhancements:
      • Crysis 3 – frame pacing improvements
      • Far Cry 3 – 3 and 4 GPU performance improvements at high quality settings, high resolution settings
      • Anno 2070 – Improved CrossFire scaling up to 34%
      • Titanfall – Resolved in game flickering with CrossFire enabled
      • Metro Last Light – Improved Crossfire scaling up to 10%
      • Eyefinity 3×1 (with three 4K panels) no longer cuts off portions of the application
      • Stuttering has been improved in certain applications when selecting mid-Eyefinity resolutions with V-sync Enabled
    • Full support for OpenGL 4.4
      • OpenGL 4.4 supports the following extensions:
        • ARB_buffer_storage
        • ARB_enhanced_layouts
        • ARB_query_buffer_object
        • ARB_clear_texture
        • ARB_texture_mirror_clamp_to_edge
        • ARB_texture_stencil8
        • ARB_vertex_type_10f_11f_11f_rev
        • ARB_multi_bind
        • ARB_bindless_texture
        • ARB_spare_texture
        • ARB_seamless_cubemap_per_texture
        • ARB_indirect_parameters
        • ARB_compute_variable_group_size
        • ARB_shader_draw_parameters
        • ARB_shader_group_vote
    • Mantle beta driver improvements:
      • BattleField 4: Performance slowdown is no longer seen when performing a task switch/Alt-tab
      • BattleField 4: Fuzzy images when playing in rotated SLS resolution with an A10 Kaveri system

Known Issues  

    • System will TDR or BSOD when encoding with Power Director 11
    • Driver installation might result in a black screen when installing on a Dual AMD Radeon R9 295X configuration under Windows 8.1 on specific platforms (see below).   The issue can be overcome by rebooting the PC; upon reboot the display driver will be installed.  The remaining Catalyst components can then be installed.
    • ASUS Crosshair V Formula-Z (990FX)
    • ASUS Maximus VI Extreme (Z87
    • ASUS Rampage IV Extreme (X79)

Download

http://support.amd.com/en-us/kb-articles/Pages/latest-catalyst-windows-beta.aspx

Display not waking up after being turned off overnight

So, another rant time. My PC is turned on pretty much 24/7, including when I’m sleeping. I set the music to play on low volume, sometimes I run BOINC client or something, so basically system is running. But since I don’t need display, I always turn it off. In the morning, when I wake up, try to turn the display on and then  it’s rage mode on.

Display refuses to wake. Smashing keyboard and mice, nothing. Music is playing normal, Caps and Num lock responds on keyboard and I can also control music with keyboard multimedia controls. Only way to get to the system was to reset it.

So, I checked on The Internet and found out there are LOADS of people with such problems. Mostly seems to be related to Windows 7, Windows 8 in combination with AMD Radeon graphic card (usually HD7xxx series). I didn’t think of it because with Windows 7 I never had such problems, only when I switched to Windows 8.1. I’m using AMD Radeon HD7950 3GB with AMD Catalyst 13.12, so that’s where similarities with other reports started to make sense. Supposedly this problem was happening before, AMD fixed it with some Catalyst version and now with Catalyst 13.12 WHQL, it seems to be back again.

What’s even more annoying is the difficulty of reproducing this issue. It’s not enough to keep PC running and turn off the display by hand, it actually has to be over night. Like AMD drivers are checking that display is turned off over night and it fucks up everything with that. Because when I turn off display through the day, I never had this problem, even though the conditions were identical and the display was turned off for the same period of time (~6 hours). So, 6 hours over day and the problem never happened to me. Have it turned of for 6 hours over night (between midnight and 7 AM), pretty much every single morning, the damn thing won’t wake up when I move the mice around.

I’ve tried everything, changing USB devices sleeping, changed CPU C states, disabled ULPC for graphic card, checked BIOS settings, nothing.

Symptoms:

Display won’t wake up after being turned off over night with PC actually running and not being in a sleep state. Display is set to turn itself of after PC is idle for 15 minutes. Keyboard and mouse seem to be responsive, but Windows refuses to wake up the display. Pressing ON/OFF button on display doesn’t make any difference, changing input mode (between VGA and DVI) using display built-in settings (or button) doesn’t make any difference. It just stays in sleep mode.

This only happens if display is turned off over night for ~6 hours. It never happened when PC was in the same exact conditions over day. Seems like AMD throws the graphic card into some sort of deeper sleep state if it’s night time based on local time read from user computer. And since it’s happening mostly on AMD HD7xxx series which have ZeroCore feature, it seems to me like AMD fucked it up pretty badly.

Temporal solution:

What seems to work for me is to unplug DVI cable from graphic card and plug it back in. This seems to re-initialize the connection and the desktop instantly appears on the screen. Same should apply to HDMI and DisplayPort connections.

Some say switching to lockscreen in Windows 8 also wakes up the display, but I haven’t tested that yet.

Permanent solution:

Well have to wait for AMD to fix this damn thing. Anyone with friends at AMD who could speed things up? Because it’s really annoying and widespread issue with no real solution other than the above.