seems they running slower.

Also i says it RT V1.0.4.0 support, but yet not implemented.

Didn´t know where i can get the newest VKruntimes...

---

Nvidia has 364.47 with Vulkan released , sofar i can see, dont use it.
Dropped cpu performance , against 356.45.

Driver failures.

Before 364.47

http://setiathome.berkeley.edu/result.php?resultid=4770344360

After:

http://setiathome.berkeley.edu/result.php?resultid=4776908346

Only one sample, but got no more time, i must change the drivers version before my GT 640 gives up.

With 356.45 , no issues. XD

---

Bad news:

https://developer.nvidia.com/vulkan-driver

The old drivers are gone, there only the WHQL drivers left.

The WHQL drivers have a 25% loss in cpu time.

At this time i got no trustable download source to tell.

---

sorry, no drivers to download at this point, they have been exchanged by the newest WHQL drivers.

---

https://vulkan.lunarg.com/signin

A new Vulkan SDK V1.0.5.0 could be downloaded. Hove theres some performance gain.

---

http://drivers.softpedia.com/downloadTag/NVIDIA%20GeForce%20Vulkan%20356.45%20Beta

Half trustable is that download location, have fun.

---

Ten days since WHQL drivers release, and they don´t recognize that there is a loss of 25% cpu time while cunshing, they don´t wana change it. Unbelievable.

---

https://software.intel.com/en-us/blogs/2016/03/14/new-intel-vulkan-beta-1540204404-graphics-driver-for-windows-78110-1540

Vulkan drivers for intel gpus 4th to 6th generation have been released, almost on 11 March 2016.

http://support.amd.com/en-us/kb-articles/Pages/radeon-vulkan-beta.aspx

There are the Radeon HD 7xxx cards and newer Vulkan drivers from AMD. Not to forget them. ;)

---

March 28th, and there are new drivers to test, i am short back...

---

So far as i can see, there is no gain on gpu AND cpu with the newest WHQL drivers yet, for this project, others are differend.

Back to 356.45 i advice.

---

At next , i would test the lunarics apps plus the new WHQL drivers, BUT download servers are down, longer than one day !?

---

no gain, back to vanilla 356.45 drivers...

---

ha, my god, 8th April ,they released a newer drivers version, always testing, will in short be back..

I´am so sorry, the drivers development drifting away from opimizing
for cruching, there must be other prioritys.

NO GAIN on the newest drivers release under Windows. (Installed the Vulkan-Runtimes too).

Before 356.45
http://setiathome.berkeley.edu/result.php?resultid=4844436284

Newest 361.91 + Runtimes
http://setiathome.berkeley.edu/result.php?resultid=4847550593

http://setiathome.berkeley.edu/result.php?resultid=4847542186

I don´t know what a have to tell about the results, there´s more testing
needed.

---

I don't get it, what are you doing here? :?
The Vulkan API has absolutely nothing to do with SETI crunching at all!
Do you test the latency of these drivers? Well, then ok...

---

Aloha, Uli

I don't get it, what are you doing here? :?
The Vulkan API has absolutely nothing to do with SETI crunching at all!
Do you test the latency of these drivers? Well, then ok...

Exactly. Until we have a Vulkan app it is only testing the current CUDA/OpenCL portions of the driver.

From what, little, I've read about Vulkan it may only have an advantage over OpenCL in terms of performance. Maybe it could lead to a single universal GPU app across all hardware vendors?

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Maybe it could lead to a single universal GPU app across all hardware vendors?

Very doubtful. OpenCL already "could". In reality there are too many vendor-specific nuances to get speedy-enough common build. Common build possible... but un-optimized one.

From what, little, I've read about Vulkan it may only have an advantage over OpenCL in terms of performance. Maybe it could lead to a single universal GPU app across all hardware vendors?

I think our buddy Jason has been working towards that goal....
But, given the immense differences between the hardware and optimizations available, I think he has taken on a yeoman's task.
And perhaps impossible to fully achieve.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Maybe it could lead to a single universal GPU app across all hardware vendors?

Very doubtful. OpenCL already "could". In reality there are too many vendor-specific nuances to get speedy-enough common build. Common build possible... but un-optimized one.

I think this is best explained by http://www.xkcd.com

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

I think this is best explained by http://www.xkcd.com

*twothumbsup*
xkcd is always good for a very good laugh! X)

---

Aloha, Uli

Don`t get me wrong, i think standards are good.
OTOH a standard app would never be as fast as a specific app.
Engineers of Vendors have different opinions how to build hardware.
Not as bad in my point of view.

---

With each crime and every kindness we birth our future.