P4 & Hyper Threading

Due to some conversation I thought I would run a quike test against HT enabled and dissabled on my P4 3GHz computer. Using a track I have been writing recently in Renoise I compared it’s results for HT both on and off. Here’s my findings for anybody interested:

With HT enabled I had a reading of 11% when first loading my track, which peaked at 53% from using the inbuilt Renoise meters. Checking this against the Task Manager performance gave 34% of the one processor, which is about 68% of what was available. This isn’t as much difference as I susspected, as I thought Renoise would give me the result of total, i.e 50% would be at maximum.

With HT dissabled I had a reading of 9.6% with the tune idle (sorry only did the idle measurements with the inbuilt Renoise meter) abd peak while playing was around 65% on both it’s and tak Manager’s reading.

So to sumerise, the real differences seem very small, with HT dissabled being very slightly ahead, but dissabled also helping Renoise to give a more accurate reading of how much CPU you really have left to use.

Anybody else done much testing? I have heard mixed results as to whether it’s worth turning of HT is using hosts that don’t suppert it, so I thought I would put my own up here.

imho if Renoise would support the HT feature there would be a bigger difference … HT is not a overall performance boost on all applications … it speeds up the multitasking of windows in general

Some also claim that it reduces your processor when it’s not being utalised, this seems to point to it making very little difference…

Should do some tests with other hosts…[SQL]

Renoise disables HT and (if I remember correctly) always assigns itself to logical CPU 0. I think it will say so in the log files as well. It was changed in v1.5 due to massive VST problems.

Here’s an old bug thread that covers some more details:

yes, but this setting can be changed in the task manager.

Using HT gives me lots of troubles in stability.

Ah yeah, that’s true. I just assumed kazakore was talking about disabling it in the BIOS (or in the driver settings or something like that). In that case, there should be no difference, other than that background processes could run on CPU1 and possibly give Renoise some extra juice.

Anyway, what I’ve learnt is that the CPU meter in Renoise is not too trustworthy - with or without HT.

Sorry yeah my fault, I was talking about dissabling in BIOS, to make you computer act as a classic, single core CPU, which some people claim will greatly improve performance when using programs that can not take advantage of the virtual second CPU.

Hope this clears up the confusion ;)