My existing GPU (NVIDIA GeForce GT 730 2GB DDR3) died and I need to replace it. I'm coming off a low knowledge base with regards to GPUs. I don't want to send too much on the GPUs. I'm not a 'Gamer' and only just want a low budget GPU that will play nice with VS. I can either get an like for like replacement (GT 730) or for a few more dollars step up to GT 1030 (2GB GDDR5). Am I heading in the right direction? I'm told that GT730 is ancient.
What should I be looking for in an GPU? What are the key aspects with regards using it with VS? What are the 'must haves' ? (I assume CUDA ?)
A friend offered to sell me his Radeon used RX 5500 XT (4GB GDDR6) for $AUD 100 but that doesn't seem support CUDA. Does that matter? Is it a good offer?
I would appreciate any feedback, thanks.
Graphics Card Advice
Moderator: Ken Berry
-
RickMen
- Posts: 258
- Joined: Mon Mar 30, 2009 6:03 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: Gigabyte B660M GAMING AC DDR4
- processor: Intel i7-12700F
- ram: 16GB
- Video Card: NVIDIA GeForce GT 730
- sound_card: Realtek HD Audio
- Hard_Drive_Capacity: 500GB M2 PCIe NVMe SDD
- Monitor/Display Make & Model: Philips 246V5 24inch wide LCD
- Corel programs: VS2022 Ultimate, VS2018 Ultimate
- Location: Sydney, Australia
Graphics Card Advice
Sony FDR-AX53, JVC Everio GZ-HD40
-
Alipod
- Posts: 9
- Joined: Fri Dec 29, 2023 7:06 pm
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: Aorus Master X570
- processor: AMD Ryzen 3700X
- ram: 32GB 3200
- Video Card: Sapphire Nitro+ AMD RX 5700 XT 8GB
- sound_card: Realtic
- Hard_Drive_Capacity: 4TB NVMe SSD
- Monitor/Display Make & Model: AGON 2K
- Corel programs: VideoStudio Ultimate 2023
Re: Graphics Card Advice
I can't give an advice about which card to get but read this link to expand your knowledge about 5000 AMD series
from wikipedia about AMD VCE
t.ly/1wFh_
from wikipedia about AMD VCE
t.ly/1wFh_
-
Vdanny
- Posts: 90
- Joined: Wed Nov 12, 2014 5:02 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- ram: 16GB
- Video Card: Nvidia 1060 6GB
Re: Graphics Card Advice
Hi,
The GT1030 is only a 64bit card so I wouldn't recommend it. If you can, I would go for a used GTX1050i or new GTX1650 which are 128bit cards and will provide a good performance boost for little money. Is your PC case a SFF one that requires a low profile card? You also need to check your PSU to see what wattage it is to see what cards you might be limited to or if you need to consider replacement.
The GT1030 is only a 64bit card so I wouldn't recommend it. If you can, I would go for a used GTX1050i or new GTX1650 which are 128bit cards and will provide a good performance boost for little money. Is your PC case a SFF one that requires a low profile card? You also need to check your PSU to see what wattage it is to see what cards you might be limited to or if you need to consider replacement.
-
RickMen
- Posts: 258
- Joined: Mon Mar 30, 2009 6:03 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: Gigabyte B660M GAMING AC DDR4
- processor: Intel i7-12700F
- ram: 16GB
- Video Card: NVIDIA GeForce GT 730
- sound_card: Realtek HD Audio
- Hard_Drive_Capacity: 500GB M2 PCIe NVMe SDD
- Monitor/Display Make & Model: Philips 246V5 24inch wide LCD
- Corel programs: VS2022 Ultimate, VS2018 Ultimate
- Location: Sydney, Australia
Re: Graphics Card Advice
My friend has installed the RX 5500 XT onto my Desktop so that I can test out the card. I thought I would run a series of renders (Share->Start using Same as Project Settings where project properties is matched first clip) under various scenarios (using a 15 sec mp4 100 Mbs single clip in the Project) under 3 conditions:
1)clean clip;
2), same clip with FX added;
3) same clip with stabilisation added),
I selected "AMD" as my Hardware type. With Hardware Encode not ticked the results were consistent, CPU hits 100% and GPU was low <10% and times were about 1:1 (ie 15 sec clip took 15-18 sec to render) BUT when I turned on Hardware Encode option I got wildly different results each time I ran the test. Sometimes the CPU was low (<10%) and GPU maxed out (>90-100%), next time (with no changes made) the CPU/GPU would be 30/20% then the next time the CPU/GPU would be 30/5% and this seems to take longer to render. That's weird, I could not consistently replicate the results.
Thanks Alipod. Food for thought. With my old GPU the rendered clips' Mbs would range from 85-90. With limited testing on the 5500 XT I'm getting 89-90 Mbs rendered clips.
Vdanny, thanks for pointing me in the right direction. If I go the nVidia route I will investigate further your recommendations. My case is a mini tower and PSU is 500W.
1)clean clip;
2), same clip with FX added;
3) same clip with stabilisation added),
I selected "AMD" as my Hardware type. With Hardware Encode not ticked the results were consistent, CPU hits 100% and GPU was low <10% and times were about 1:1 (ie 15 sec clip took 15-18 sec to render) BUT when I turned on Hardware Encode option I got wildly different results each time I ran the test. Sometimes the CPU was low (<10%) and GPU maxed out (>90-100%), next time (with no changes made) the CPU/GPU would be 30/20% then the next time the CPU/GPU would be 30/5% and this seems to take longer to render. That's weird, I could not consistently replicate the results.
Thanks Alipod. Food for thought. With my old GPU the rendered clips' Mbs would range from 85-90. With limited testing on the 5500 XT I'm getting 89-90 Mbs rendered clips.
Vdanny, thanks for pointing me in the right direction. If I go the nVidia route I will investigate further your recommendations. My case is a mini tower and PSU is 500W.
Sony FDR-AX53, JVC Everio GZ-HD40
-
RickMen
- Posts: 258
- Joined: Mon Mar 30, 2009 6:03 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: Gigabyte B660M GAMING AC DDR4
- processor: Intel i7-12700F
- ram: 16GB
- Video Card: NVIDIA GeForce GT 730
- sound_card: Realtek HD Audio
- Hard_Drive_Capacity: 500GB M2 PCIe NVMe SDD
- Monitor/Display Make & Model: Philips 246V5 24inch wide LCD
- Corel programs: VS2022 Ultimate, VS2018 Ultimate
- Location: Sydney, Australia
Re: Graphics Card Advice
The FX clip I used above was 'Lens Flare'. When I used another FX the results were more consistent. I suspect the Len Flare FX is buggy.
Sony FDR-AX53, JVC Everio GZ-HD40
-
RickMen
- Posts: 258
- Joined: Mon Mar 30, 2009 6:03 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: Gigabyte B660M GAMING AC DDR4
- processor: Intel i7-12700F
- ram: 16GB
- Video Card: NVIDIA GeForce GT 730
- sound_card: Realtek HD Audio
- Hard_Drive_Capacity: 500GB M2 PCIe NVMe SDD
- Monitor/Display Make & Model: Philips 246V5 24inch wide LCD
- Corel programs: VS2022 Ultimate, VS2018 Ultimate
- Location: Sydney, Australia
Re: Graphics Card Advice
So I re-ran a series of renders (Share->Start using Same as Project Settings where project properties is matched to the first clip using a 15 sec mp4 100 Mb/s single clip in the Project, Smart Render was On) under 3 conditions:
1)clean clip;
2)same clip with FX added; (of course actual results vary depending on which FX used)
3) same clip with stabilisation added,
With Hardware Encode not ticked the results were consistent, CPU hits 100% and GPU was about 10% and times were about 1:2 (ie 15 sec clip took 27-31 sec to render for pts 2 & 3). With Hardware Encode option selected CPU / GPU was around 30%/30% and render times closer 1:1. Under both settings (Hardware Encode On and Off) the rendered files were about 90 Mb/s.
Conclusion - the 5500XT produced faster rendering times compared to just the CPU, there was not a drop off in Mb/s. 5500XT plays nice with VS and it's massive improvement over my old GT730 performance.
I then stressed tested (Furmark) the 5500XT, no apparent issues.
So either $100 for a used 5500XT or $200-300 for a new GTX1650. The 5500XT ranks above the GTX 1650. I'm leaning towards the AMD 5500XT.
1)clean clip;
2)same clip with FX added; (of course actual results vary depending on which FX used)
3) same clip with stabilisation added,
With Hardware Encode not ticked the results were consistent, CPU hits 100% and GPU was about 10% and times were about 1:2 (ie 15 sec clip took 27-31 sec to render for pts 2 & 3). With Hardware Encode option selected CPU / GPU was around 30%/30% and render times closer 1:1. Under both settings (Hardware Encode On and Off) the rendered files were about 90 Mb/s.
Conclusion - the 5500XT produced faster rendering times compared to just the CPU, there was not a drop off in Mb/s. 5500XT plays nice with VS and it's massive improvement over my old GT730 performance.
I then stressed tested (Furmark) the 5500XT, no apparent issues.
So either $100 for a used 5500XT or $200-300 for a new GTX1650. The 5500XT ranks above the GTX 1650. I'm leaning towards the AMD 5500XT.
Sony FDR-AX53, JVC Everio GZ-HD40
-
Vdanny
- Posts: 90
- Joined: Wed Nov 12, 2014 5:02 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- ram: 16GB
- Video Card: Nvidia 1060 6GB
Re: Graphics Card Advice
Hi,
That is great that you have a beefy power supply that can accommodate the Radeon. Considering the price difference and the performance gain you can already see I would also probably go for the RX 5500 XT despite its lack of CUDA.
That is great that you have a beefy power supply that can accommodate the Radeon. Considering the price difference and the performance gain you can already see I would also probably go for the RX 5500 XT despite its lack of CUDA.
-
gewb
- Posts: 467
- Joined: Wed Dec 02, 2015 3:29 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: MSI Pro B550-VC
- processor: AMD Ryzen 7 5700X
- ram: 16GB
- Video Card: ASRock Radeon RX 6650 XT
- Hard_Drive_Capacity: 4TB
- Monitor/Display Make & Model: LG 24MK430
- Corel programs: 2018, 2021, 2023 Ult.
- Location: Colorado, USA
- Contact:
Re: Graphics Card Advice
The main parts of CUDA are the ISA called PTX which is what NVCC and other CUDA compilers target and the CUDA runtime which is the virtual machine that can run PTX binaries on compatible hardware. Thus, AMD does not use CUDA. There are several initiatives to translate/cross-compile CUDA to different languages and APIs but it isn't going anywhere fast.
