BM3D (VapourSynth) versus MVTools - Page 3

WolframRhodium · 1st August 2021, 12:26

Quote:

Originally Posted by tormento

Does Test2 work for CUDA too or is it AVX2 bonded?

Both are included and are implemented in different dlls.

real.finder · 1st August 2021, 12:36

Code:

ColorBars(width=640, height=480, pixel_type="yv12")
LoadPlugin("BM3DCPU_AVS.dll")
ConvertBits(bits=32)
BM3D_cpu(sigma=0.5, radius=2)
BM3D_VAggregate(radius=2)
ConvertBits(8)

22 fps with no mt (no Prefetch) on i7-8700

I have a question about BM3D_VAggregate, it's cpu only? and seems it's Duplicate in both cpu and cuda dlls and that seems not ok

WolframRhodium · 1st August 2021, 13:20

Quote:

Originally Posted by real.finder

Code:

ColorBars(width=640, height=480, pixel_type="yv12")
LoadPlugin("BM3DCPU_AVS.dll")
ConvertBits(bits=32)
BM3D_cpu(sigma=0.5, radius=2)
BM3D_VAggregate(radius=2)
ConvertBits(8)

22 fps with no mt (no Prefetch) on i7-8700

I have a question about BM3D_VAggregate, it's cpu only? and seems it's Duplicate in both cpu and cuda dlls and that seems not ok

It's cpu only since its performance is limited by memory bandwidth and has low arithmetic intensity. Test2 moved that function to the cpu dll so there should be no conflict.

tormento · 2nd August 2021, 09:25

Quote:

Originally Posted by WolframRhodium

Both are included and are implemented in different dlls.

Any changelog for test2 on CUDA?

kedautinh12 · 2nd August 2021, 09:32

Read readme file in file downloaded

WolframRhodium · 2nd August 2021, 11:29

Quote:

Originally Posted by tormento

Any changelog for test2 on CUDA?

Nothing besides now it depends on the CPU version on VAggregate.

Reel.Deel · 3rd August 2021, 00:58

Quote:

Originally Posted by WolframRhodium

Nothing besides now it depends on the CPU version on VAggregate.

Any possibility of also supporting AVX? I see the VS version supports it, according to the documentation: The cpu version is implemented in AVX/AVX2 intrinsics...

kedautinh12 · 3rd August 2021, 01:48

Quote:

Originally Posted by Reel.Deel

Any possibility of also supporting AVX? I see the VS version supports it, according to the documentation: The cpu version is implemented in AVX/AVX2 intrinsics...

But in Requirements, he wrote only AVX2

WolframRhodium · 3rd August 2021, 02:08

I mean AVX and AVX2 intrinsics. Sorry for the confusion.

Reel.Deel · 3rd August 2021, 02:44

Quote:

Originally Posted by WolframRhodium

I mean AVX and AVX2 intrinsics. Sorry for the confusion.

Ahh gotcha, I guess that explains this error in AVSMeter:

Code:

Exception: 0xC000001D [STATUS_ILLEGAL_INSTRUCTION]
Module: BM3DCPU_AVS.dll
Address: 0x000007FEC45204E6

Figured it was worth asking about AVX since I see just "AVX" in the requirements of the VS version, but in other parts it says AVX/AVX2.... well I guess that leaves me out of it till I build my next PC

tormento · 5th August 2021, 20:24

Quote:

Originally Posted by WolframRhodium

I mean AVX and AVX2 intrinsics. Sorry for the confusion.

I am having AVS+ exception error when using test2 on the following script:

SetMemoryMax()
SetFilterMTMode("DEFAULT_MT_MODE", 2)
LoadPlugin("D:\Eseguibili\Media\DGDecNV\DGDecodeNV.dll")
DGSource("F:\In\1_33 Grosso guaio a Chinatown\grosso.dgi",ct=132,cb=132,cl=0,cr=0)
ConvertBits(bits=32)
BM3D_cuda(sigma=0.5, radius=2)
BM3D_VAggregate(radius=2)
ConvertBits(16)
Prefetch(6)

I am using a SandyBridge with no AVX2 but it should give no error with the CUDA version. Am I doing something wrong?

Reel.Deel · 5th August 2021, 20:39

Quote:

Originally Posted by tormento

I am using a SandyBridge with no AVX2 but it should give no error with the CUDA version. Am I doing something wrong?

"BM3D_CUDA" depends on "BM3D_CPU" for temporal denoising. In other words, avx2 capable processor is required.

@WolframRhodium

Since the first test version worked on the gpu only, would it be possible to add a parameter to force gpu processing only? For those that do not have avx2.

tormento · 5th August 2021, 20:41

Quote:

Originally Posted by Reel.Deel

Since the first test version worked on the gpu only, would it be possible to add a parameter to force gpu processing only? For those that do not have avx2.

I think it would be easier to provide 2 versions as every script author should otherwise check if cpu supports avx2 and I don't know if there is such capability in AVS+.

kedautinh12 · 5th August 2021, 23:41

Quote:

Originally Posted by tormento

I am having AVS+ exception error when using test2 on the following script:

SetMemoryMax()
SetFilterMTMode("DEFAULT_MT_MODE", 2)
LoadPlugin("D:\Eseguibili\Media\DGDecNV\DGDecodeNV.dll")
DGSource("F:\In\1_33 Grosso guaio a Chinatown\grosso.dgi",ct=132,cb=132,cl=0,cr=0)
ConvertBits(bits=32)
BM3D_cuda(sigma=0.5, radius=2)
BM3D_VAggregate(radius=2)
ConvertBits(16)
Prefetch(6)

I am using a SandyBridge with no AVX2 but it should give no error with the CUDA version. Am I doing something wrong?

Of course BM3D_VAggregate() move to BM3DCPU in test2, try delete BM3D_VAggregate() and check script again

WolframRhodium · 6th August 2021, 06:51

Thanks for the report. Please check the latest version (test3). I separate VAggregate from BM3D_CPU.

tormento · 6th August 2021, 10:30

Quote:

Originally Posted by kedautinh12

Of course BM3D_VAggregate() move to BM3DCPU in test2, try delete BM3D_VAggregate() and check script again

No offense, I know you try to help, but sometimes I feel your posts are just to increase post counting.

How can you run BM3D without aggregation?

tormento · 6th August 2021, 10:32

Quote:

Originally Posted by WolframRhodium

Thanks for the report. Please check the latest version (test3). I separate VAggregate from BM3D_CPU.

Thanks, I will try ASAP and report.

kedautinh12 · 6th August 2021, 10:53

Quote:

Originally Posted by tormento

No offense, I know you try to help, but sometimes I feel your posts are just to increase post counting.

How can you run BM3D without aggregation?

Lol, i know i'm noob. I just try help with my ability

tormento · 6th August 2021, 13:34

Quote:

Originally Posted by WolframRhodium

Thanks for the report. Please check the latest version (test3). I separate VAggregate from BM3D_CPU.

Tested and working.

Now, I have to understand which is better between this and KNLMeansCL...

Reel.Deel · 9th August 2021, 08:35

Quote:

Originally Posted by WolframRhodium

Thanks for the report. Please check the latest version (test3). I separate VAggregate from BM3D_CPU.

Thank for the update. I also confirm that it works

Question, are arrays allowed? I took a peek at the source code (line 429) and it made me think that arrays are allowed just like the VS version.
When I set sigma to [0.5, 0.0, 0.5] it sets the U plane to 255, if I use [0.5, 0.0, 0.0] it sets both U/V to 255. If I use [0.0,0.5,0.5] it sets the Y plane to 255. So I guess that using an array does work, I just don't know if setting sigma to 0 should behave that way.

1st August 2021, 12:36	#42 \| Link
real.finder Registered User Join Date: Jan 2012 Location: Mesopotamia Posts: 2,587	Code: ColorBars(width=640, height=480, pixel_type="yv12") LoadPlugin("BM3DCPU_AVS.dll") ConvertBits(bits=32) BM3D_cpu(sigma=0.5, radius=2) BM3D_VAggregate(radius=2) ConvertBits(8) 22 fps with no mt (no Prefetch) on i7-8700 I have a question about BM3D_VAggregate, it's cpu only? and seems it's Duplicate in both cpu and cuda dlls and that seems not ok __________________ See My Avisynth Stuff

2nd August 2021, 09:32	#45 \| Link
kedautinh12 Registered User Join Date: Jan 2018 Posts: 2,156	Read readme file in file downloaded

3rd August 2021, 02:08	#49 \| Link
WolframRhodium Registered User Join Date: Jan 2016 Posts: 162	I mean AVX and AVX2 intrinsics. Sorry for the confusion.

6th August 2021, 06:51	#55 \| Link
WolframRhodium Registered User Join Date: Jan 2016 Posts: 162	Thanks for the report. Please check the latest version (test3). I separate VAggregate from BM3D_CPU.