I have Dell Inspiron i7-5500 cpu @2.4ghz cpu with 16gb RAM, pretty basic.
I don't want to spend tons of money, want to know how can I increase the throughput.
Is it simple to get 1 or 2 GPUs and hook it up or how best to upgrade my current hardware.
Thanks, Raghu
First you have to unhide your computers so people can see what you have right now, it's hard to make suggestions based on not knowing what you have to start with.
The easiest way to increase your throughput is a GPU. They do more work than a CPU in the same amount of time. However, you have a dell computer. They have limited power supply and usually don't support an high end graphic card. I used to have a dell as well. It had a single PCIe slot (where an additional GPU could go) but didn't have any extra power supply cables to power an additional graphic card. What worked best a low end GPU that got all it's power from the GPU slot. In my case it was a nvidia 750 but you could probably get a 1050ti to run in your case as long as you have room. Don't bother trying to add a more powerful power supply, the motherboards are coded in such a way that they don't support larger PSU or high end GPUs.
The easiest way to increase your throughput is a GPU. They do more work than a CPU in the same amount of time. However, you have a dell computer. They have limited power supply and usually don't support an high end graphic card. I used to have a dell as well. It had a single PCIe slot (where an additional GPU could go) but didn't have any extra power supply cables to power an additional graphic card. What worked best a low end GPU that got all it's power from the GPU slot. In my case it was a nvidia 750 but you could probably get a 1050ti to run in your case as long as you have room. Don't bother trying to add a more powerful power supply, the motherboards are coded in such a way that they don't support larger PSU or high end GPUs.
I just bought two refurbed Dell's that I was able to get an adapter from the Dell 8 pin power supply plug to a 750 watt standard 24 pin power supply plug and can now run a 1060 in each of them. They are i7-4770 cpu's, I DID have to remove most of the drive cage though to make the longer gpu card fit in them, but since they are just crunchers that doesn't bother me. Now two of my older Dell's can only take a 750Ti as the max gpu even though I also put a 750 watt psu in them too, something is hard coded somewhere and newer gpu's just won't work, they have I7-3930 cpu's in them. As you say though Dell's are finicky machines to upgrade!!!
Thanks a lot Zalster and Mikey for your response! I'll see if I can upgrade with a GPU. If I stay motivated for a year or 2, will then upgrade to a High End setup :)
A 1050ti or RX 560 with no 6-pin power connectors are great options. The 1050 ti is probably a little better GPU than the RX 560, but their around the same price and both are 75 watts from the PCI-express.
I started this thread a while ago, but I've been running SETI for the past couple months and wish to return to running Einstein as well. I have a RX570 and a RX 470, with an i5-4590 CPU on windows 10 by the way.
How many tasks would you recommend running per GPU? What should I do to change these preferences? Thanks!
[quote=Joshua a RX570 and a RX 470, with an i5-4590 CPU on windows 10 by the way.
How many tasks would you recommend running per GPU? What should I do to change these preferences? Thanks!
You could try if 2 tasks per card would run without problems. That would be pretty much optimal. Make the first task start running and then wait about 4 minutes until you let the second task start in parallel (this step for each card). That will lead to a good situation where the final stage of computation in those tasks won't happen at the same time --> GPU will have something to do at all times which will be a good thing.
You could adjust the settings (Account - Preferences - Project) by changing the "GPU utilization factor of FGRP apps:" to 0.50 and save. Make sure to adjust the preferences of the same location/venue that your computer is using.
Or you could instead save a app_config.xml into your Einstein project folder (which you can find under your Boinc data folder / projects). The app_config.xml could have these lines:
I followed your advice and used the account preferences method to adjust the settings (Account - Preferences - Project) by changing the "GPU utilization factor of FGRP apps:" to 0.50 and save. I'm now using 2 tasks per card. I was finishing tasks in around 10 minutes with 1 task per card. I'll update with how it does with 2 tasks.
raghuram wrote:Hi, I'm new
)
First you have to unhide your computers so people can see what you have right now, it's hard to make suggestions based on not knowing what you have to start with.
Thanks Mikey, I've now made
)
Thanks Mikey, I've now made the change, wasn't aware.
The easiest way to increase
)
The easiest way to increase your throughput is a GPU. They do more work than a CPU in the same amount of time. However, you have a dell computer. They have limited power supply and usually don't support an high end graphic card. I used to have a dell as well. It had a single PCIe slot (where an additional GPU could go) but didn't have any extra power supply cables to power an additional graphic card. What worked best a low end GPU that got all it's power from the GPU slot. In my case it was a nvidia 750 but you could probably get a 1050ti to run in your case as long as you have room. Don't bother trying to add a more powerful power supply, the motherboards are coded in such a way that they don't support larger PSU or high end GPUs.
Zalster wrote:The easiest way
)
I just bought two refurbed Dell's that I was able to get an adapter from the Dell 8 pin power supply plug to a 750 watt standard 24 pin power supply plug and can now run a 1060 in each of them. They are i7-4770 cpu's, I DID have to remove most of the drive cage though to make the longer gpu card fit in them, but since they are just crunchers that doesn't bother me. Now two of my older Dell's can only take a 750Ti as the max gpu even though I also put a 750 watt psu in them too, something is hard coded somewhere and newer gpu's just won't work, they have I7-3930 cpu's in them. As you say though Dell's are finicky machines to upgrade!!!
Thanks a lot Zalster and
)
Thanks a lot Zalster and Mikey for your response! I'll see if I can upgrade with a GPU. If I stay motivated for a year or 2, will then upgrade to a High End setup :)
A 1050ti or RX 560 with no
)
A 1050ti or RX 560 with no 6-pin power connectors are great options. The 1050 ti is probably a little better GPU than the RX 560, but their around the same price and both are 75 watts from the PCI-express.
Thanks a lot Joshua! Will
)
Thanks a lot Joshua! Will check
I started this thread a while
)
I started this thread a while ago, but I've been running SETI for the past couple months and wish to return to running Einstein as well. I have a RX570 and a RX 470, with an i5-4590 CPU on windows 10 by the way.
How many tasks would you recommend running per GPU? What should I do to change these preferences? Thanks!
[quote=Joshua a RX570 and a
)
[quote=Joshua a RX570 and a RX 470, with an i5-4590 CPU on windows 10 by the way.
How many tasks would you recommend running per GPU? What should I do to change these preferences? Thanks!
You could try if 2 tasks per card would run without problems. That would be pretty much optimal. Make the first task start running and then wait about 4 minutes until you let the second task start in parallel (this step for each card). That will lead to a good situation where the final stage of computation in those tasks won't happen at the same time --> GPU will have something to do at all times which will be a good thing.
You could adjust the settings (Account - Preferences - Project) by changing the "GPU utilization factor of FGRP apps:" to 0.50 and save. Make sure to adjust the preferences of the same location/venue that your computer is using.
Or you could instead save a app_config.xml into your Einstein project folder (which you can find under your Boinc data folder / projects). The app_config.xml could have these lines:
<app_config>
<app>
<name>hsgamma_FGRPB1G</name>
<gpu_versions>
<gpu_usage>.5</gpu_usage>
<cpu_usage>1</cpu_usage>
</gpu_versions>
</app>
</app_config>
Which ever way you choose... force 'update' for Einstein in Boinc. Then the settings should start effecting more or less soon.
edit: I'm not sure again if the gpu_usage 0.5 is right for system with 2 cards, but other people will correct...
I followed your advice and
)
I followed your advice and used the account preferences method to adjust the settings (Account - Preferences - Project) by changing the "GPU utilization factor of FGRP apps:" to 0.50 and save. I'm now using 2 tasks per card. I was finishing tasks in around 10 minutes with 1 task per card. I'll update with how it does with 2 tasks.
Thanks!