Project 2665: Uploads are 2X the size of the data file?

[Foxery]

I've seen a quirk with two different 2665 WUs now, that I haven't noticed on other clients/projects. (I'll have to look more carefully, though) This is with Windows SMP v5.92, under Windows XP.
Project: 2665 (Run 1, Clone 774, Gen 4)
Project: 2665 (Run 3, Clone 261, Gen 4)

My connection monitor shows that I am uploading double the amount of data than the file size. e.g. the client reports:
[01:25:20] + Attempting to send results
[01:25:20] - Reading file work/wuresults_02.dat from core
[01:25:20] (Read 22075772 bytes from disk)

I can see in the \Work directory that yes, the results file is 22MB. However, my software firewall keeps track of how much I upload/download, and it just incremented by 44MB. These WUs have both been accepted by the work server, and I've been creditted the points. Weird, eh?

[Foxery]

/Bump for some love...

Bandwidth isn't a big concern for me, but certainly could be for some donors.

[toTOW]

I never noticed that ... are you sure that you didn't have a WU waiting in queue ?

[Foxery]

Ach. I finally got something other than P2665 or P2653, and was around to see it upload. This double-bandwidth issue also exists for p3065.

If I'm not mistaken, it takes people on slow DSL (128kbits upstream) half an hour to send results for P2665 - actually a full hour with this bug. Ouch.

[nwkelley]

hey foxery, i'll shoot the SMP guys a line to see if there is anything abnormal here that they know of... thanks for your keen eye...

[VijayPande]

This one is a new one. On the surface, I can't think of why this would happen. Could you see if this happens again?

[toTOW]

Just a note : I suspect the same behavior with the GPU client ... I asked Dan to check the WU returned, but I don't know if he had the time to do so ...

[^w^ing]

using netlimiter, i've observed many 2665's results being uploaded and all of them uploaded the exact amount of data it should, according to the fahlog (the

+ Attempting to send results
- Reading file work/wuresults_0x.dat from core
(Read xxxxxxx bytes from disk)

part)
for 2665s it was around 20MBs if i remember it correctly.

[Foxery]

This one is a new one. On the surface, I can't think of why this would happen. Could you see if this happens again?

It has with every p2665 WU I have done on this machine - the most noticeable and most troubling one, as it has very large (22MB) result files. I don't think it occurs on the trusty old p2653 WUs. My Windows machine hasn't gotten anything else in ages for me to check.

toTOW: I do not believe this occurs on the GPU client, but its results are so tiny... I'll try to watch a few tonight just to be sure.