[CptAsian]

In the next few weeks, I'll be setting up a folding rig with a pair of Fury cards. I plan to run it on Linux and possibly dual-boot Windows as well, so I'd be more than happy post a comparison if I get the chance to.

 

[CptAsian]

Just submitted a pair of R9 Fury WUs @ stock clocks. I plan to submit a few more over the week with varying clocks.

 

[CptAsian]

I've been submitting a few more pairs of R9 Fury work units, some at stock clocks (1050 MHz) and some OC'd (1110 MHz). Might do a few more if I can remember.

 

[CptAsian]

Quote:

Originally Posted by tictoc

Just submitted another unit for my RX 480 in Linux. It looks like the Linux performance has now caught up to Windows (and maybe even passed it), at least for Polaris.

April 2017 Result
p9415 - 220k PPD
October 2017 Result
p9415 - 295k PPD

@CptAsian
@msgclb
do you guys have any recent results for p9415 on your AMD cards?

I'm just going to list out my PPD for my last 10 P9415 WUs from the month of October. I didn't include those that seem to be affected by pausing for over an hour or so. Note that some numbers may be skewed because I do other stuff (light gaming, Youtube, Twitch, etc.) while I fold, which I imagine impacts PPD.
These are from two identical cards on Windows 10. Sapphire R9 Furys, undervolted at stock clocks, so 1050 MHz.

251K
244K
231K
244K
240K
237K
250K
238K
247K
241K

Average is 242.3K PPD.

I hope that's helpful!

 

[CptAsian]

Since I saw you guys talking about the 1050 Ti in another thread, I was inspired to add 10 WUs from my 960: five older ones and my five most recent ones.

 

[CptAsian]

hey guys - i just checked in on folding here... and the [email protected] server is not receiving any WUs?
Seems like it is affecting everyone.

Yep, I just checked the official folding forums and it looks like people are expecting things to come back tomorrow, being a weekday.

 

[CptAsian]

Good evening folks!

I have been working with Blade over the past few weeks on redoing the PPD Database, and we have pushed it live today!

The following changes have been made to the Database:

• Revamped the back-end - Improved front-end loading time and performance, automated the addition of new hardware and projects to the database.
• Removed antiquated Yield calculations - theoretical number calculated has been deprecated now that GPUs are significantly faster.
• Replaced individual submission list with averages - Increase front-end performance.
• Added Project# and OS tabs - Provides averages and sample size for each P# per GPU and OS/GPU submitted to the database.
• Changed GPU Sort Order - The GTX 10xx GPUs really should not have been listed before the GTX 4xx series GPUs.
• Cleaned up formatting -
• Laid out framework for accepting bulk submissions - See below for more details.

We have been working on this a while now, but we are willing to take suggestions for any additions or changes you think may be beneficial.

In regards to bulk submissions, we are working on a system that will allow users to submit properly formatted .csv files from HFM. We have a few issues we are working out before it's ready for release, but more to come on that. We will also need beta testers at some point, so I will reach out to you all when we are ready for testing. The individual submission form still works, and we encourage you to keep using it!

Anyways, more to come soon! :thumb:

Nice to see things are being kept fresh in this thread.


However, if I'm really being honest, I liked that big, unwieldy list of individual submissions. It made it really easy to tell if there were outliers, if most of the submissions came from one user with a highly-OC'd or (probably) watercooled card, and so on. The OS sheet isn't too bad to use instead, and I can see why that big list was removed. Just my two cents.


Bulk submissions sound great! The only thing is users who fold on non-24/7 rigs will have to be careful about submitting WUs that were paused, but that's not the end of the world. Definitely worth having that feature.

 

[CptAsian]

Wow, I didn't realize newer cards had made such a jump in folding performance, even in the sort of mid-range of the lineup. My 1070 Tis stock make around 750-800K PPD and they're only a little bit behind on gaming performance and were only slightly cheaper when I got them on clearance. Interesting.

 

[CptAsian]

Was just checking some numbers and saw that the 1660 has one submission and it's with a PPD value of 542, and that can't be right. Is it possible to manually delete/manipulate that data somewhere?

 

[CptAsian]

Re-pinned.

 

[CptAsian]

@Simmons572 looks like the rogue GTX 1660 entry is back and I'm doubtful of the 1660S entries as well.

 

[CptAsian]

I am aware. I was tempted not to delete them as we have a sample size of 1 and 2 respectively. But at the same time, that data is very incorrect, and we need some WU submissions for those two GPUs. If somebody just so happens to have these cards...

There's a small chance I'll have my hands on a 1660S for a little while in the coming months, which is why I checked/noticed those. If that works out I'll be sure to submit some results.

 

[CptAsian]

Perfect. Once we get some consistent WU's collected, we can get those erroneous units removed. Also, make sure you are tracking with HFM. I am still willing to process bulk submissions for people using HFM.

Yep, I'm planning quite a few hardware shuffles and OS reinstalls, so before that I can probably bulk submit some 1070 Ti/980 Ti/980/960 units. I'll start that PM eventually, hopefully.

 

[CptAsian]

Something I just thought of: do we have submission dates tied to the individual entries? If so, I think it would be useful to have something like an "average submission date" that would give some indication to whether or not the average PPD for a card is recent, since that can change over time due to driver updates or something like the recent CUDA update.

 

[CptAsian]

I feel like that could get a bit messy. Average date might work, but I will need to investigate.. Does anyone else like this idea?

Yeah I'm sure there's a better way to do it but that's the only thing that's come to mind so far.