[Phoronix] Radeon ROCm 2.0 Officially Out With OpenCL 2.0 Support - Overclock.net - An Overclocking Community

Forum Jump: 

[Phoronix] Radeon ROCm 2.0 Officially Out With OpenCL 2.0 Support

 
Thread Tools
Old 12-20-2018, 12:20 PM - Thread Starter
sudo apt install sl
WannaBeOCer is an unknown quantity at this point WannaBeOCer is an unknown quantity at this point
 
WannaBeOCer's Avatar
 
Join Date: Dec 2009
Posts: 3,802
[Phoronix] Radeon ROCm 2.0 Officially Out With OpenCL 2.0 Support

Source: https://www.phoronix.com/scan.php?pa...Cm-2.0-Arrives

ROCm on Github: https://github.com/RadeonOpenCompute/ROCm

Edit: Tensorflow-ROCm https://github.com/ROCmSoftwarePlatf...rflow-upstream

Quote:
AMD reached their goal of delivering the feature-packed ROCm 2.0 in 2018. Yesterday I covered the primary highlights on this big Radeon Open Compute stack update when there were signs of ROCm 2.0 being prepared for release this week. That milestone has now been officially released with ROCm 2.0 now being available, including the RHEL/CentOS and Ubuntu ROCm 2.0 binaries for easy installation.
Quote:
New features and enhancements in ROCm 2.0
Adds support for RHEL 7.6 / CentOS 7.6 and Ubuntu 18.04.1
Adds support for Vega 7nm
Introduces MIVisionX
A comprehensive computer vision and machine intelligence libraries, utilities and applications bundled into a single toolkit.
Improvements to ROCm Libraries
rocSPARSE & hipSPARSE
rocBLAS with improved DGEMM efficiency on Vega 7nm
MIOpen
This release contains general bug fixes and an updated performance database
Group convolutions backwards weights performance has been improved
RNNs now support fp16
Tensorflow multi-gpu and Tensorflow FP16 support for Vega 7nm
TensorFlow v1.12 is enabled with fp16 support
PyTorch/Caffe2 with Vega 7nm Support
fp16 support is enabled
Several bug fixes and performance enhancements
Known Issue: breaking changes are introduced in ROCm 2.0 which are not addressed upstream yet. Meanwhile, please continue to use ROCm fork at https://github.com/ROCmSoftwarePlatform/pytorch
Improvements to ROCProfiler tool
Support for Vega 7nm
Support for hipStreamCreateWithPriority
Creates a stream with the specified priority. It creates a stream on which enqueued kernels have a different priority for execution compared to kernels enqueued on normal priority streams. The priority could be higher or lower than normal priority streams.
OpenCL 2.0 support
ROCm 2.0 introduces full support for kernels written in the OpenCL 2.0 C language on certain devices and systems. Applications can detect this support by calling the “clGetDeviceInfo” query function with “parame_name” argument set to “CL_DEVICE_OPENCL_C_VERSION”. In order to make use of OpenCL 2.0 C language features, the application must include the option “-cl-std=CL2.0” in options passed to the runtime API calls responsible for compiling or building device programs. The complete specification for the OpenCL 2.0 C language can be obtained using the following link: https://www.khronos.org/registry/Ope....0-openclc.pdf
Improved Virtual Addressing (48 bit VA) management for Vega 10 and later GPUs
Fixes Clang AddressSanitizer and potentially other 3rd-party memory debugging tools with ROCm
Small performance improvement on workloads that do a lot of memory management
Removes virtual address space limitations on systems with more VRAM than system memory
Kubernetes support

Maximus
(18 items)
CPU
Core i7 6700K 4.8Ghz @ 1.4v
Motherboard
Maximus VIII Formula
GPU
Radeon RX VEGA 64 @ 1750/1105Mhz with 1175Mv
RAM
G-Skill 32GB 3200Mhz
Hard Drive
Samsung 850 Evo 2TB Raid 0
Power Supply
EVGA SuperNova 1200w P2
Cooling
EK Supremacy Full Copper Clean
Cooling
EK-FC Radeon Vega
Cooling
XSPC D5 Photon v2
Cooling
Black Ice Gen 2 GTX360 x2
Case
Thermaltake Core X5 Tempered Glass Edition
Operating System
Windows 10 Pro
Monitor
Acer XF270HUA
Keyboard
Cherry MX Board 6.0
Mouse
Logitech G600
Mouse
Alugraphics GamerArt
Audio
Definitive Technology Incline
Audio
SMSL M8A
▲ hide details ▲

Last edited by WannaBeOCer; 12-20-2018 at 12:25 PM.
WannaBeOCer is offline  
Sponsored Links
Advertisement
 
Old 12-20-2018, 04:48 PM
Sunday League Jibber
SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about SuperZan has a spectacular aura about
 
SuperZan's Avatar
 
Join Date: Sep 2015
Location: Latveria
Posts: 4,243
I've been very much looking forward to this.

Farewel happy Fields; Where Joy for ever dwells; Hail horrours, hail; Infernal world, and thou profoundest Hell; Receive thy new Possessor; One who brings; A mind not to be chang'd by Place or Time; The mind is its own place, and in it self; Can make a Heav'n of Hell, a Hell of Heav'n.
SuperZan's Doombot
(14 items)
SuperZan's YT-1300
(14 items)
CPU
AMD Ryzen 7 2700X
Motherboard
ASRock X370 Taichi
GPU
Zotac GTX 1080 Ti Leather Jacket Edition 11GB
RAM
G.SKILL TridentZ RGB Series 16GB (2 x 8GB) @ 3200 14-14-14-34 1T
Hard Drive
Samsung 850 Pro 256GB SSD (Windows)
Hard Drive
SanDisk Ultra II 480GB SSD (Games)
Hard Drive
Seagate 2TB SSHD (Storage)
Hard Drive
Intel 530 Series 480GB SSD (Linux)
Optical Drive
LG DVD RW
Power Supply
EVGA Supernova G2 850w
Cooling
EK-Supremacy MX CPU Block
Cooling
EK-XRES 100 + DDC Pump
Cooling
XSPC Razor (1080 Ti) GPU Block
Cooling
EK-CoolStream PE 360 Rad
Cooling
XSPC EX280 Rad
Case
Corsair 760T
Operating System
Windows 10 Pro
Monitor
Pixio PX277 27" 1440p 144Hz DP-Adaptive Sync w/ LFC
Keyboard
JAR-OWL J1 Rainbow LED Mechanical
Mouse
Zowie FK1
Mousepad
Razor Firefly
Audio
Plantronics Rig 500e Gaming Headset
Audio
Marantz Pro MPH-1 Monitoring Headphones
Audio
S.M.S.L SD-793II Audio DAC + Headphone Amplifier
Other
Xbox One Controller
Other
Wacom Intuous S Drawing Pad (CTL4100)
CPU
AMD Ryzen R3 1200
Motherboard
ASRock AB350m Pro4
GPU
PowerColor PCS+ R9 380x 4GB
RAM
Samsung 8 GB (2x4GB) @ 2400
Hard Drive
Biostar M200-240 240GB m.2 SATA SSD (Solus)
Hard Drive
Kingston V200 128GB SSD (Windows)
Hard Drive
WD Blue 250GB HDD
Power Supply
Seasonic S12II 430 Bronze 430w
Cooling
AMD Prism RGB Cooler
Case
DIYPC Cuboid-G Black/Green
Operating System
Solus Budgie
Operating System
Windows 10 Pro
Keyboard
LYILIN 2.4GHz Mini Wireless & Bluetooth Keyboard with Touchpad Mouse
Other
Steam Controller
CPU
Intel Core i7-4790k @ 4.6GHz
Motherboard
Asrock Z97 Extreme6
GPU
EVGA GTX 1070 SC
RAM
Corsair Vengeance 16GB (2x8GB) @ 2133 10-10-10-28 1T
Hard Drive
Samsung 860 Evo 1TB
Optical Drive
LG DVD RW
Power Supply
EVGA SuperNOVA NEX 650w
Cooling
Cooler Master Hyper 212 Plus
Case
Corsair Carbide 300R
Operating System
Windows 10
Monitor
40" Vizio HDTV
Keyboard
HP Generic
Mouse
Logitech Generic
Audio
HyperX Cloud
▲ hide details ▲


SuperZan is online now  
Old 12-21-2018, 12:18 AM
Top kek
ku4eto is an unknown quantity at this point ku4eto is an unknown quantity at this point
 
Join Date: Oct 2013
Location: Bulgaria , Sofia
Posts: 3,285
You said, you will test Tensorflow performance with your Vega....

Still waiting.

Previous Hardware:
Spoiler!
Desktop PC
(19 items)
CPU
AMD FX-8320
Motherboard
ASRock Fatal1ty 990FX Killer
GPU
Sapphire Nitro+ RX480
GPU
XFX RX470 Singlefan
GPU
MSI RX580 GamingX 4GB
RAM
Mushkin Redline 996996 2x4GB 2133Mhz
Hard Drive
Western Digital 160GB 7200RPM 8MB
Hard Drive
Maxtor 250GB 7200RPM 8MB
Hard Drive
Corsair Force LS
Hard Drive
WesternDigital Blue 500GB 7200RPM 16MB
Power Supply
Corsair TX850M
Cooling
ThermalTake Frio Silent 14
Case
ThermalTake View 27
Operating System
Windows 10 Enterprise 1607
Operating System
Linux Mint 17.3 Rosa
Monitor
AOC i2267FWH
Keyboard
Logitech K120
Mouse
Bloody V5
Audio
Corsair HS30 Raptor
▲ hide details ▲
ku4eto is offline  
Sponsored Links
Advertisement
 
Old 12-21-2018, 07:02 PM - Thread Starter
sudo apt install sl
WannaBeOCer is an unknown quantity at this point WannaBeOCer is an unknown quantity at this point
 
WannaBeOCer's Avatar
 
Join Date: Dec 2009
Posts: 3,802
Quote: Originally Posted by ku4eto View Post
You said, you will test Tensorflow performance with your Vega....

Still waiting.
Wanted to run the benchmarks after I upgraded.

Setup:

Ubuntu 18.04.1
amdgpu-pro 18.50
ROCm Version: 2.0.89
Tensorflow-ROCm 1.12

AlexNet 32

Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 679.5 +/- 0.0 (jitter = 0.0) nan
10 images/sec: 799.0 +/- 36.8 (jitter = 31.1) nan
20 images/sec: 773.5 +/- 26.6 (jitter = 171.8) nan
30 images/sec: 782.4 +/- 21.5 (jitter = 33.8) nan
40 images/sec: 770.9 +/- 19.2 (jitter = 162.1) nan
50 images/sec: 777.1 +/- 17.0 (jitter = 37.4) nan
60 images/sec: 770.2 +/- 15.7 (jitter = 162.7) nan
70 images/sec: 769.5 +/- 14.7 (jitter = 165.3) nan
80 images/sec: 774.3 +/- 13.6 (jitter = 42.7) nan
90 images/sec: 770.1 +/- 12.9 (jitter = 161.2) nan
100 images/sec: 767.6 +/- 12.3 (jitter = 176.3) nan
----------------------------------------------------------------
total images/sec: 764.88
----------------------------------------------------------------
AlexNet batch size 64

Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 947.0 +/- 0.0 (jitter = 0.0) 7.200
10 images/sec: 979.8 +/- 26.3 (jitter = 69.6) 7.200
20 images/sec: 982.8 +/- 18.1 (jitter = 74.2) 7.200
30 images/sec: 995.7 +/- 15.6 (jitter = 110.8) 7.200
40 images/sec: 992.4 +/- 13.1 (jitter = 111.0) 7.200
50 images/sec: 993.5 +/- 11.7 (jitter = 114.1) 7.200
60 images/sec: 990.0 +/- 10.5 (jitter = 95.8) 7.200
70 images/sec: 979.3 +/- 10.0 (jitter = 81.4) 7.199
80 images/sec: 982.8 +/- 9.4 (jitter = 95.0) 7.200
90 images/sec: 989.2 +/- 9.0 (jitter = 108.1) 7.200
100 images/sec: 988.2 +/- 8.4 (jitter = 101.4) 7.200
----------------------------------------------------------------
total images/sec: 986.01
----------------------------------------------------------------
GoogleNet batch size 32

Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 396.2 +/- 0.0 (jitter = 0.0) 7.186
10 images/sec: 355.9 +/- 13.0 (jitter = 47.8) 7.099
20 images/sec: 366.3 +/- 8.0 (jitter = 3.5) 7.096
30 images/sec: 364.5 +/- 6.4 (jitter = 7.5) 7.072
40 images/sec: 365.4 +/- 5.7 (jitter = 4.8) 7.113
50 images/sec: 363.8 +/- 5.1 (jitter = 7.0) 7.047
60 images/sec: 363.6 +/- 4.7 (jitter = 7.3) 7.121
70 images/sec: 364.4 +/- 4.3 (jitter = 7.4) 7.099
80 images/sec: 364.7 +/- 4.0 (jitter = 7.7) 7.099
90 images/sec: 365.8 +/- 3.7 (jitter = 5.6) 7.088
100 images/sec: 364.8 +/- 3.5 (jitter = 6.9) 7.105
----------------------------------------------------------------
total images/sec: 364.46
----------------------------------------------------------------
GoogleNet batch size 64

Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 348.6 +/- 0.0 (jitter = 0.0) 7.078
10 images/sec: 380.2 +/- 6.6 (jitter = 3.1) 7.044
20 images/sec: 379.9 +/- 4.9 (jitter = 3.3) 7.101
30 images/sec: 381.5 +/- 3.4 (jitter = 2.8) 7.097
40 images/sec: 382.4 +/- 3.3 (jitter = 2.6) 7.116
50 images/sec: 382.5 +/- 2.8 (jitter = 2.9) 7.100
60 images/sec: 381.6 +/- 2.5 (jitter = 2.9) 7.084
70 images/sec: 381.8 +/- 2.2 (jitter = 2.7) 7.081
80 images/sec: 382.1 +/- 2.1 (jitter = 3.1) 7.097
90 images/sec: 383.6 +/- 2.1 (jitter = 3.0) 7.102
100 images/sec: 383.2 +/- 1.9 (jitter = 2.8) 7.077
----------------------------------------------------------------
total images/sec: 382.89
----------------------------------------------------------------
ResNet50 batch size 32
Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 151.0 +/- 0.0 (jitter = 0.0) 8.458
10 images/sec: 149.0 +/- 1.4 (jitter = 0.9) 7.997
20 images/sec: 147.8 +/- 1.1 (jitter = 2.0) 8.260
30 images/sec: 148.0 +/- 1.0 (jitter = 2.5) 8.339
40 images/sec: 148.8 +/- 0.9 (jitter = 1.9) 8.187
50 images/sec: 149.1 +/- 0.8 (jitter = 2.5) 7.754
60 images/sec: 148.7 +/- 0.9 (jitter = 3.8) 8.077
70 images/sec: 148.8 +/- 0.8 (jitter = 4.3) 8.481
80 images/sec: 148.3 +/- 0.8 (jitter = 3.3) 8.286
90 images/sec: 148.4 +/- 0.7 (jitter = 3.2) 8.027
100 images/sec: 148.6 +/- 0.7 (jitter = 2.8) 8.008
----------------------------------------------------------------
total images/sec: 148.52
----------------------------------------------------------------
ResNet batch size 64
Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 168.4 +/- 0.0 (jitter = 0.0) 8.218
10 images/sec: 170.4 +/- 1.9 (jitter = 4.3) 8.124
20 images/sec: 169.4 +/- 1.2 (jitter = 5.7) 8.228
30 images/sec: 170.3 +/- 1.0 (jitter = 7.0) 8.265
40 images/sec: 170.2 +/- 0.9 (jitter = 7.0) 8.354
50 images/sec: 169.7 +/- 0.8 (jitter = 4.8) 8.017
60 images/sec: 169.7 +/- 0.7 (jitter = 4.0) 8.271
70 images/sec: 169.6 +/- 0.6 (jitter = 4.3) 8.305
80 images/sec: 169.7 +/- 0.6 (jitter = 4.2) 8.212
90 images/sec: 170.0 +/- 0.5 (jitter = 4.3) 8.309
100 images/sec: 169.9 +/- 0.5 (jitter = 4.5) 8.279
----------------------------------------------------------------
total images/sec: 169.79
----------------------------------------------------------------

Edit: Here's Resnet 50 with batch size 64 with my RX Vega 64 set to P7 with memory set to 945Mhz. The top benchmarks were at stock.

Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 183.8 +/- 0.0 (jitter = 0.0) 8.217
10 images/sec: 184.2 +/- 0.3 (jitter = 0.6) 8.124
20 images/sec: 184.0 +/- 0.2 (jitter = 0.8) 8.226
30 images/sec: 184.1 +/- 0.1 (jitter = 0.7) 8.267
40 images/sec: 184.1 +/- 0.2 (jitter = 0.9) 8.345
50 images/sec: 184.1 +/- 0.1 (jitter = 0.9) 8.013
60 images/sec: 184.0 +/- 0.2 (jitter = 0.9) 8.264
70 images/sec: 183.9 +/- 0.2 (jitter = 0.9) 8.322
80 images/sec: 183.8 +/- 0.2 (jitter = 0.8) 8.209
90 images/sec: 183.8 +/- 0.2 (jitter = 0.8) 8.278
100 images/sec: 183.9 +/- 0.2 (jitter = 0.8) 8.277
----------------------------------------------------------------
total images/sec: 183.79
---------------------------------------------------------------
ResNet 50 batch size 64 using FP16

Quote:
Done warm up
Step Img/sec total_loss
1 images/sec: 194.7 +/- 0.0 (jitter = 0.0) 8.156
10 images/sec: 194.5 +/- 0.9 (jitter = 1.0) 8.200
20 images/sec: 194.1 +/- 0.8 (jitter = 0.7) 8.388
30 images/sec: 194.4 +/- 0.6 (jitter = 1.2) 8.273
40 images/sec: 194.5 +/- 0.4 (jitter = 0.9) 8.135
50 images/sec: 194.4 +/- 0.4 (jitter = 1.0) 8.166
60 images/sec: 194.6 +/- 0.4 (jitter = 1.0) 8.323
70 images/sec: 194.6 +/- 0.3 (jitter = 0.9) 8.375
80 images/sec: 194.7 +/- 0.3 (jitter = 0.9) 8.010
90 images/sec: 194.5 +/- 0.3 (jitter = 0.9) 8.293
100 images/sec: 194.6 +/- 0.3 (jitter = 0.9) 8.349
----------------------------------------------------------------
total images/sec: 194.47
----------------------------------------------------------------
Attached Thumbnails
Click image for larger version

Name:	Screenshot from 2018-12-21 18-04-24.png
Views:	28
Size:	501.6 KB
ID:	240318  


Maximus
(18 items)
CPU
Core i7 6700K 4.8Ghz @ 1.4v
Motherboard
Maximus VIII Formula
GPU
Radeon RX VEGA 64 @ 1750/1105Mhz with 1175Mv
RAM
G-Skill 32GB 3200Mhz
Hard Drive
Samsung 850 Evo 2TB Raid 0
Power Supply
EVGA SuperNova 1200w P2
Cooling
EK Supremacy Full Copper Clean
Cooling
EK-FC Radeon Vega
Cooling
XSPC D5 Photon v2
Cooling
Black Ice Gen 2 GTX360 x2
Case
Thermaltake Core X5 Tempered Glass Edition
Operating System
Windows 10 Pro
Monitor
Acer XF270HUA
Keyboard
Cherry MX Board 6.0
Mouse
Logitech G600
Mouse
Alugraphics GamerArt
Audio
Definitive Technology Incline
Audio
SMSL M8A
▲ hide details ▲

Last edited by WannaBeOCer; 12-21-2018 at 07:22 PM.
WannaBeOCer is offline  
Old 12-22-2018, 02:56 AM
Top kek
ku4eto is an unknown quantity at this point ku4eto is an unknown quantity at this point
 
Join Date: Oct 2013
Location: Bulgaria , Sofia
Posts: 3,285
Can you tell which you used? Looking at the Phoronix benchmarks:


https://www.phoronix.com/scan.php?pa...nsorflow&num=1

Not sure if its tested the same way.

Previous Hardware:
Spoiler!
Desktop PC
(19 items)
CPU
AMD FX-8320
Motherboard
ASRock Fatal1ty 990FX Killer
GPU
Sapphire Nitro+ RX480
GPU
XFX RX470 Singlefan
GPU
MSI RX580 GamingX 4GB
RAM
Mushkin Redline 996996 2x4GB 2133Mhz
Hard Drive
Western Digital 160GB 7200RPM 8MB
Hard Drive
Maxtor 250GB 7200RPM 8MB
Hard Drive
Corsair Force LS
Hard Drive
WesternDigital Blue 500GB 7200RPM 16MB
Power Supply
Corsair TX850M
Cooling
ThermalTake Frio Silent 14
Case
ThermalTake View 27
Operating System
Windows 10 Enterprise 1607
Operating System
Linux Mint 17.3 Rosa
Monitor
AOC i2267FWH
Keyboard
Logitech K120
Mouse
Bloody V5
Audio
Corsair HS30 Raptor
▲ hide details ▲
ku4eto is offline  
Old 12-22-2018, 02:17 PM - Thread Starter
sudo apt install sl
WannaBeOCer is an unknown quantity at this point WannaBeOCer is an unknown quantity at this point
 
WannaBeOCer's Avatar
 
Join Date: Dec 2009
Posts: 3,802
Quote: Originally Posted by ku4eto View Post
Can you tell which you used? Looking at the Phoronix benchmarks:


https://www.phoronix.com/scan.php?pa...nsorflow&num=1

Not sure if its tested the same way.
He's using FP16 along with the nVidia container. I should be able to test to run the Tensorflow benchmarks I ran on a GTX 1080 over the the holiday break.

https://docs.nvidia.com/deeplearning...rel_18.09.html


This is the benchmark I ran:

https://github.com/tensorflow/benchm...atible?files=1

Maximus
(18 items)
CPU
Core i7 6700K 4.8Ghz @ 1.4v
Motherboard
Maximus VIII Formula
GPU
Radeon RX VEGA 64 @ 1750/1105Mhz with 1175Mv
RAM
G-Skill 32GB 3200Mhz
Hard Drive
Samsung 850 Evo 2TB Raid 0
Power Supply
EVGA SuperNova 1200w P2
Cooling
EK Supremacy Full Copper Clean
Cooling
EK-FC Radeon Vega
Cooling
XSPC D5 Photon v2
Cooling
Black Ice Gen 2 GTX360 x2
Case
Thermaltake Core X5 Tempered Glass Edition
Operating System
Windows 10 Pro
Monitor
Acer XF270HUA
Keyboard
Cherry MX Board 6.0
Mouse
Logitech G600
Mouse
Alugraphics GamerArt
Audio
Definitive Technology Incline
Audio
SMSL M8A
▲ hide details ▲
WannaBeOCer is offline  
 

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in

Thread Tools


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off