Notifications
Clear all

Belkin 2m TB3 cable is a winner  

  RSS

Trajans Row
(@trajans-row)
Active Member
Joined: 5 years ago
 

After experiencing disconnect issues with a Cable Matters branded 2m Thunderbolt 40Gb cable, I decided to try a Belkin TB3 cable of the same length. It's cheaper, too - $52 on the Apple Store.

 
So far, the Belkin seems solid with my tMBP and Core. The plugs on on the cable are longer than the .5m cable from Razer, which permits more leverage on the port. However, I was pull the cable 360 degrees around the port with no disconnects during a benchmark. It also looks to be charging the laptop, so it does carry at least some power. The sheathing is pretty flexible, making organization better.
 
Would recommend. 

To do: Create my signature with system and expected eGPU configuration information to give context to my posts. I have no builds.

.

itsage liked
ReplyQuote
itsage
(@itsage)
Founder Admin
Joined: 5 years ago
 

Have you confirmed it's a 40Gb/s cable? I got a Belkin 1m Thunderbolt 20Gb/s cable. It is slightly slower (about 10%) than the AKiTiO .5m Thunderbolt 40GB/s during benchmark. It was tested with the internal display on my 13" MBP with the Node.

LinkedIn | Twitter | Facebook | Youtube | Instagram
 
external graphics card builds
best laptops for external GPU
eGPU enclosure buyer's guide

 
2021 13" ASUS ROG Flow X13 (RTX3050Ti) [R5K,8C,H] + RTX 3080 @ 64Gbps-XG (XG Mobile) + Win10 20H2 [build link]  


ikir liked
ReplyQuote
dsaul
(@dsaul)
Eminent Member
Joined: 5 years ago
 

The 0.5m and 2m cable are 40GB/s. I wanted to by 1m cable coz of the price and would be ideal for me but it is indeed 20GB/s. Not sure why they don't make the 1m cable at 40GB/s.

2016 13" MacBook Pro [6th,2C,U] + GTX 1060 @ 32Gbps-TB3 (AKiTiO Thunder3) + Win10 [build link]  

ReplyQuote
Trajans Row
(@trajans-row)
Active Member
Joined: 5 years ago
 

I'm not sure how to tell in Windows what the negotiated TB link speed is, but a SteamVR benchmark on the internal LCD gives me the same results with both cables. It is also advertised to support 40Gb, so I have to assume the cable is capable of active mode. 

To do: Create my signature with system and expected eGPU configuration information to give context to my posts. I have no builds.

.

ReplyQuote
Alec
 Alec
(@alec)
Trusted Member
Joined: 5 years ago
 

I have the same cable and get random crashes when playing games that use 100% of the GPU.

2016 15" MacBook Pro (RP450) [6th,4C,H] + GTX 1080 @ 32Gbps-TB3 (Razer Core) + Win10 [build link]  

ReplyQuote
nando4
(@nando4)
Noble Member Admin
Joined: 5 years ago
 

A bit of a contradiction to the opening post.

@tunatoro tested the Cable Matters 40Gbps 2m TB3 cable and got full TB3 performance as measured by CUDA-Z at:

https://egpu.io/forums/builds/acer-aspire-v17-nitro-black-editionakitio-nodegtx1070-win-10/

He previously tested the Cable Matters 20Gbps 2m TB3 cable and saw reduced performance in the above thread.

 

eGPU Setup 1.35    •    eGPU Port Bandwidth Reference Table

 
2015 15" Dell Precision 7510 (Q M1000M) [6th,4C,H] + GTX 1080 Ti @32Gbps-M.2 (ADT-Link R43SG) + Win10 1803 // compares M.2 vs TB3 performance inc unoptimized H-CPU BIOS [build link]  


ReplyQuote
Francesco Ripa
(@francesco_ripa)
Active Member
Joined: 10 months ago
 

A brief comparison between the cable supplied with ADT-Link and the Belkin 40Gb/s cable:

ADT-Link

Platform 0 : Intel(R) Corporation
Platform 1 : Advanced Micro Devices, Inc.
Platform found : Advanced Micro Devices, Inc.

Selected Platform Vendor : Advanced Micro Devices, Inc.
Device 0 : Ellesmere Device ID is 000001456CE9B1C0
Build Options are : -cl-std=CL2.0
Build: _WINxx release
GPU work items: 32768
Buffer size: 33554432
CPU workers: 1
Timing loops: 20
Repeats: 1
Kernel loops: 20
inputBuffer: CL_MEM_READ_ONLY
outputBuffer: CL_MEM_WRITE_ONLY
inputSVMBuffer: CL_MEM_READ_ONLY
outputSVMBuffer: CL_MEM_WRITE_ONLY

Host baseline (naive):

Timer resolution 101.72 ns
Page fault 558.17 ns
Barrier speed 84.36 ns

CPU read 10.55 GB/s
memcpy() 8.64 GB/s
memset(,1,) 25.54 GB/s
memset(,0,) 27.15 GB/s

AVERAGES (over loops 2 - 19, use -l for complete log)
--------

1. Host mapped write to inputBuffer
---------------------------------------|---------------
clEnqueueMapBuffer -- WRITE (GBPS) | 2105.193
---------------------------------------|---------------
memset() (GBPS) | 27.443
---------------------------------------|---------------
clEnqueueUnmapMemObject() (GBPS) | 2.740

2. GPU kernel read of inputBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.003

Verification Passed!

3. GPU kernel write to outputBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.005

4. Host mapped read of outputBuffer
---------------------------------------|---------------
clEnqueueMapBuffer -- READ (GBPS) | 2.857
---------------------------------------|---------------
CPU read (GBPS) | 9.720
---------------------------------------|---------------
clEnqueueUnmapMemObject() (GBPS) | 808.974

Verification Passed!

5. Host mapped write to inputSVMBuffer
---------------------------------------|---------------
clEnqueueSVMMap -- WRITE (GBPS) | 2444.273
---------------------------------------|---------------
memset() (GBPS) | 26.643
---------------------------------------|---------------
clEnqueueSVMUnmap() (GBPS) | 2.744

6. GPU kernel execution using inputSVMBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.003

Verification Passed!

7. GPU kernel write to outputSVMBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.005

8. Host mapped read of outputSVMBuffer
---------------------------------------|---------------
clEnqueueSVMMap -- READ (GBPS) | 2.852
---------------------------------------|---------------
CPU read (GBPS) | 9.463
---------------------------------------|---------------
clEnqueueSVMUnmap() (GBPS) | 920.701

Verification Passed!

Belkin

Platform 0 : Intel(R) Corporation
Platform 1 : Advanced Micro Devices, Inc.
Platform found : Advanced Micro Devices, Inc.

Selected Platform Vendor : Advanced Micro Devices, Inc.
Device 0 : Ellesmere Device ID is 000001E6FDDA1990
Build Options are : -cl-std=CL2.0
Build: _WINxx release
GPU work items: 32768
Buffer size: 33554432
CPU workers: 1
Timing loops: 20
Repeats: 1
Kernel loops: 20
inputBuffer: CL_MEM_READ_ONLY
outputBuffer: CL_MEM_WRITE_ONLY
inputSVMBuffer: CL_MEM_READ_ONLY
outputSVMBuffer: CL_MEM_WRITE_ONLY

Host baseline (naive):

Timer resolution 103.35 ns
Page fault 2167.87 ns
Barrier speed 356.73 ns

CPU read 8.63 GB/s
memcpy() 7.79 GB/s
memset(,1,) 23.62 GB/s
memset(,0,) 23.61 GB/s

AVERAGES (over loops 2 - 19, use -l for complete log)
--------

1. Host mapped write to inputBuffer
---------------------------------------|---------------
clEnqueueMapBuffer -- WRITE (GBPS) | 3359.176
---------------------------------------|---------------
memset() (GBPS) | 27.255
---------------------------------------|---------------
clEnqueueUnmapMemObject() (GBPS) | 2.728

2. GPU kernel read of inputBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.003

Verification Passed!

3. GPU kernel write to outputBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.005

4. Host mapped read of outputBuffer
---------------------------------------|---------------
clEnqueueMapBuffer -- READ (GBPS) | 2.839
---------------------------------------|---------------
CPU read (GBPS) | 9.361
---------------------------------------|---------------
clEnqueueUnmapMemObject() (GBPS) | 935.677

Verification Passed!

5. Host mapped write to inputSVMBuffer
---------------------------------------|---------------
clEnqueueSVMMap -- WRITE (GBPS) | 3108.491
---------------------------------------|---------------
memset() (GBPS) | 26.855
---------------------------------------|---------------
clEnqueueSVMUnmap() (GBPS) | 2.737

6. GPU kernel execution using inputSVMBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.003

Verification Passed!

7. GPU kernel write to outputSVMBuffer
---------------------------------------|---------------
clEnqueueNDRangeKernel() (secs) | 0.005

8. Host mapped read of outputSVMBuffer
---------------------------------------|---------------
clEnqueueSVMMap -- READ (GBPS) | 2.844
---------------------------------------|---------------
CPU read (GBPS) | 9.604
---------------------------------------|---------------
clEnqueueSVMUnmap() (GBPS) | 976.997

Verification Passed!

[Please login to see this]

2019 MacBook Air (MacBookAir8,2, 1,6 GHz, 8 GB) + RX 570 @ R43SG-TB3 + macOS 10.15.7 & Win10


ReplyQuote