RTX 2060 Folding at GTX 1070 Ti Speeds

[bruce]

[+quote="Mstenholm"]The first WU came in before I had set client-type = advanced and only netted 450k PPD. I didn't catch what project that WU came from, but I'll make a note and report out if it happens again.

The "advanced" setting doesn't change the PPD but does enable you to be assigned WU from other projects.

The title of this topic suggests that there's some kind of misunderstanding. As a general rule, NVidia GPUs that are xx70 or xx70Ti are expected to run about the same speed as (xx+1)60. To get a GPU with increased speed, you'll need to keep the next-to-the-last-digit (in this case, the 7) the same or greater (i.e. greater than 7).

[katakaio]

The "advanced" setting doesn't change the PPD but does enable you to be assigned WU from other projects.

Understood - that's why I was curious to note what project that first WU came from before I had set the advanced flag. I'd also be curious if that WU was a fluke and other 2060 users are running without the advanced flag and still seeing 1M+ PPD.

[gordonbb]

The "advanced" setting doesn't change the PPD but does enable you to be assigned WU from other projects.

Understood - that's why I was curious to note what project that first WU came from before I had set the advanced flag. I'd also be curious if that WU was a fluke and other 2060 users are running without the advanced flag and still seeing 1M+ PPD.

Yes, I’ve never bothered to set the Advanced flag as that feature is deprecated and I’m getting 900,000 to 1,100,00PPD on the 2060.

Completed 15 or so WUs so far and like the 2070 have been impressed with its power efficiency over the 1000 series. I’ve just gotten my X Display Manager working again and had to rearrange the cards in the system as the issue was gdm3 not liking a DVI to HDMI adaptor (my only free display only has VGA & HDMI). So n a week or so I should have some data in HFM.net to analyze

[Mstenholm]

[+quote="Mstenholm"]The first WU came in before I had set client-type = advanced and only netted 450k PPD. I didn't catch what project that WU came from, but I'll make a note and report out if it happens again.

The "advanced" setting doesn't change the PPD but does enable you to be assigned WU from other projects.

The title of this topic suggests that there's some kind of misunderstanding. As a general rule, NVidia GPUs that are xx70 or xx70Ti are expected to run about the same speed as (xx+1)60. To get a GPU with increased speed, you'll need to keep the next-to-the-last-digit (in this case, the 7) the same or greater (i.e. greater than 7).

I never mentioned anything about client-type = advance and 450K. Get the quotes straight please. I found that there is a small advantage to the 1070ti OC +125 MHz to a stock 2060.

11728 864K for the 1070ti 1936 MHz and 766K for 1890 MHz. The case run hot with 90% fan on the top 1070ti.

[Joe_H]

Yes, I’ve never bothered to set the Advanced flag as that feature is deprecated

What gives you the impression that the Advanced flag is deprecated? It most certainly is not deprecated. It merely indicates that a folder is willing to accept WU's from projects that have recently been moved out of beta testing. They might still have some issues with hardware and OS configurations not encountered during initial testing.

[Nathan_P]

Current very early testing with my Zotac 2060 base model. Its currently running a 14131 WU at 1m PPD, using 150w front the wall. Clocks are boosting to 1875mhz on Mint 18.2

Missing out i maybe but the current cards are running well and I want a cpu based system next - going to bring back some good old fashioned heavy duty cpu folding. my current dual 12c ivy bridge xeons just don't cut it.

[gordonbb]

The title of this topic suggests that there's some kind of misunderstanding. As a general rule, NVidia GPUs that are xx70 or xx70Ti are expected to run about the same speed as (xx+1)60. To get a GPU with increased speed, you'll need to keep the next-to-the-last-digit (in this case, the 7) the same or greater (i.e. greater than 7).

Agreed that there is a general perception that, based on previous launches, the 2060 SHOULD have be expected to perform similar to a 1070. Now if the price parity between series had been maintained ...

The original title, however, was meant to convey based on the initial SP FAHBench results that the 2060 was, in fact, closer in performance to a 1070Ti and thus punching way above it's weight class as a 1070 Ti is generally accepted to be able to perform almost the same as a 1080 and thus the 2060 represents a very good value for Folding.

Yes, I’ve never bothered to set the Advanced flag as that feature is deprecated

What gives you the impression that the Advanced flag is deprecated? It most certainly is not deprecated. It merely indicates that a folder is willing to accept WU's from projects that have recently been moved out of beta testing. They might still have some issues with hardware and OS configurations not encountered during initial testing.

Ahh OK, I was thinking of the Big Advanced flag not the Advanced flag.

Once I get things tuned a bit better I will set that on my Development system that has a 1060 6GB in it. I assume that once the WUs are out of Beta that there's really no point reporting WUs that consistently run at a lower PPD like 11718?

[gordonbb]

... I found that there is a small advantage to the 1070ti OC +125 MHz to a stock 2060.

11728 864K for the 1070ti 1936 MHz and 766K for 1890 MHz. The case run hot with 90% fan on the top 1070ti.

I assume you mean:

WU Model Clock PPD
11728 1070 Ti 1936 864k
11728 2060 1890 766k

I'm seeing:

WU GPU Clock PPD
11728 1070 Ti 2012 900-922kPPD
11728 2060 1974 1.1-1.2MPPD

The 1070Ti is in a lower slot of a dual GPU system and was running at an average of 162W, 52% Fan and 65.85C
The 2060 was in the upper slot of another dual GPU system and was running at an average of 155.4W, 62.24% Fan and 71.39C

[bruce]

Ahh OK, I was thinking of the Big Advanced flag not the Advanced flag.

The "big advanced" flag has been deprecated. There are no longer any projects associated with that configuration setting.

[Theodore]

The title of this topic suggests that there's some kind of misunderstanding. As a general rule, NVidia GPUs that are xx70 or xx70Ti are expected to run about the same speed as (xx+1)60. To get a GPU with increased speed, you'll need to keep the next-to-the-last-digit (in this case, the 7) the same or greater (i.e. greater than 7).

Agreed that there is a general perception that, based on previous launches, the 2060 SHOULD have be expected to perform similar to a 1070. Now if the price parity between series had been maintained ...

The original title, however, was meant to convey based on the initial SP FAHBench results that the 2060 was, in fact, closer in performance to a 1070Ti and thus punching way above it's weight class as a 1070 Ti is generally accepted to be able to perform almost the same as a 1080 and thus the 2060 represents a very good value for Folding.

Of the 2060, there seem to be a few models out.
The base model, and some call an 'amp' model, or 'gaming' model with higher overclock, and supposedly premium components.
The major premium component being the Samsung VRAM vs Micron VRAM, which Samsung runs more stable at higher overclock speeds.
It appears that some models of the 2060 run at 1400Mhz, while others run at 1800+Mhz. That's a big difference!

[JimF]

It appears that some models of the 2060 run at 1400Mhz, while others run at 1800+Mhz. That's a big difference!

It may be that they are quoting different scenarios. For example, for the "ASUS ROG STRIX GeForce RTX 2060 DirectX 12 ROG-STRIX-RTX2060-O6G-GAMING 6GB" as shown on Newegg, they cite:

Core Clock OC Mode: 1395 MHz
Gaming Mode: 1365 MHz
Boost Clock OC Mode: 1860 MHz
Gaming Mode: 1830 MHz

What does that mean? What do you compare with what?

[gordonbb]

Of the 2060, there seem to be a few models out.
The base model, and some call an 'amp' model, or 'gaming' model with higher overclock, and supposedly premium components.
The major premium component being the Samsung VRAM vs Hynix VRAM, which Samsung runs more stable at higher overclock speeds.
It appears that some models of the 2060 run at 1400Mhz, while others run at 1800+Mhz. That's a big difference!

The 2060 from NVidia is specified to run at a base clock of 1365 and a boost of 1680MHz. If you scroll down a little on the list and look at the EVGA RTX 2060 XC Ultra it is, again specified with a 1365MHz base clock but the boost clock is specified as 1830MHz so this just tells us that EVGA’s voltage curve for this model is more aggressive than Nvidia’s likely based on their cooler design and possibly a beefier power delivery stage.

In the end it all comes down to thermals and the silicon lottery. If you have low temperatures you can clock these easily into the 1900MHz range and likely up to the low 2000s. So the specified boost clock can be adjusted upwards with a GPU clock offset and the card with a lower boost rating will require a higher offset to get to the same speed. So the boost clock rating is just a indicator of what the card can deliver in stock configuration and a hint at what the card might be capable if the Silicon does not limit performance.

Memory on Nvidia cards doesn’t really matter for folding. You will see little, if any, increase in PPD by Overclocking memory nor will you see additional PPD by underclocking memory to add thermal headroom for the GPU.

[Theodore]

View this post:
https://forums.geforce.com/default/topi ... g-memory-/

Not all cards have Samsung memory, but it seems to be a good move from Nvidia!

[Mstenholm]

... I found that there is a small advantage to the 1070ti OC +125 MHz to a stock 2060.

11728 864K for the 1070ti 1936 MHz and 766K for 1890 MHz. The case run hot with 90% fan on the top 1070ti.

I assume you mean:

WU Model Clock PPD
11728 1070 Ti 1936 864k
11728 2060 1890 766k

I'm seeing:

WU GPU Clock PPD
11728 1070 Ti 2012 900-922kPPD
11728 2060 1974 1.1-1.2MPPD

The 1070Ti is in a lower slot of a dual GPU system and was running at an average of 162W, 52% Fan and 65.85C
The 2060 was in the upper slot of another dual GPU system and was running at an average of 155.4W, 62.24% Fan and 71.39C

Yes that was I meant at the time. It turned out that I mixed up the numbers/GPUs. I got a few more and now correct numbers in this docs.google https://docs.google.com/spreadsheets/d/ ... edit#gid=0
I assume that you run with a faster CPU than I do since 1M PPD @2025 core MHz is that best I have seen so far on the 2060 and 790k on the 1070ti @ 1966 MHz core. More data in the sheet. The 2060 is clearly a better folding card running colder and using less energy.

[gordonbb]

... I assume that you run with a faster CPU than I do since 1M PPD @2025 core MHz is that best I have seen so far on the 2060 and 790k on the 1070ti @ 1966 MHz core. More data in the sheet. The 2060 is clearly a better folding card running colder and using less energy.

I don’t know about having a faster CPU for the 2060. I’m currently driving it on a GigaByte Aorus z370 Gaming 5 at PCIe3 x8 with a Pentium Gold 5500 and a single stick of DDR4-2400 along with a 2070 in another x8 slot. I’ve run the 2070 with similar results in a 10 year old Acer mini ITX board with 3GB of DDR3 and a Athlon II x2 220 at 2.8GHz in a PCIe2 x16 so I think the case thermals are more of a limiting factor than a CPU as long as a single core/thread is reserved for feeding the GPU.

There’s another thread in this sub form where the consensus seems to be that a CPU won’t limit a mid to high-tier card too much as long as it’s at least 2GHz or so.

The 1070Ti is in another rig with a GigaByte z370 SLI with a Pentium Gold 5400 and the other stick of DDR4-2400 along with a 1060 6GB.

I’ve now dropped the power limit on the cards and am measuring their efficiency. The results so far over a 3 day average are:

2060 1.04MPPD 140W 1.93GHz 69C
2070 1.29MPPD 160W 1.94GHz 63C
1070Ti 885kPPD 145W 1.96GHz 64C
1060 6G 447kPPD 110W 1.96GHz 73C