I would not currently recommend an Entanet based ISP (which I believe ADSL24 are) due to problems over the last few months which are why I moved away. They used to be the canine's family jewels so they might sort themselves out again in time but there are still problems - go look around there forum on TBB for more detail.
As others have pointed out, location will definitely affect the results of latency tests, and time will probably do so too (though far more so with some ISPs than others).
Edit: what follows isn't overly relevant as I've just seen your note that you can't get LLU providers at your exchange...
Here are my results on Be (O2 LLU should give the same results, but not O2 "Access" which people currently report congestion issues), located in York, with default packet size, gaming mode turned off:
rtr:~# ping -c 8 jolt.co.uk
PING jolt.co.uk (84.234.17.86) 56(84) bytes of data.
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=1 ttl=125 time=32.3 ms
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=2 ttl=125 time=32.8 ms
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=3 ttl=125 time=33.1 ms
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=4 ttl=125 time=32.6 ms
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=5 ttl=125 time=32.7 ms
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=6 ttl=125 time=32.7 ms
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=7 ttl=125 time=33.0 ms
64 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=8 ttl=125 time=32.8 ms
--- jolt.co.uk ping statistics ---
8 packets transmitted, 8 received, 0% packet loss, time 7006ms
rtt min/avg/max/mdev = 32.323/32.778/33.123/0.224 ms
These times are very consistent no matter what time of day I check, and are pretty much the same as I got from Entanet at good times (though even at their best they were a little more variable than I find Be to be).
See this graph for ping tests to the BBC over the last day and
this one for the last few hours - the big blip after noon yesterday was while the connection was busy downloading a pile of big stuff, and around midnight was a fairly active interactive browsing session(youtube and such). The little blip just before 0940 was while the Debian/Lenny net-inst iso downloaded.
I turned gaming mode on once just for a quick test, and it shaved more then 8ms off those times. I didn't record the data so I can't say more accurately than that. I switched it back off again as I saw a drop in upstream throughput and I'm not much of an online gamer so the throughput is more useful to me than the 8+ms latency benefit. Some people report a better saving, sometimes in the 20ms range, buy I'm guessing that they had higher latency to start with.
Some tests with larger packet sizes (this will be a more realistic test of game performance, as a game protocol is unlikely to be sending 56byte packets most of the time):
1Kbyte packets:
rtr:~# ping -c 4 -s 1024 jolt.co.uk
PING jolt.co.uk (84.234.17.86) 1024(1052) bytes of data.
1032 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=1 ttl=125 time=40.4 ms
1032 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=2 ttl=125 time=39.7 ms
1032 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=3 ttl=125 time=40.0 ms
1032 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=4 ttl=125 time=39.9 ms
--- jolt.co.uk ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3008ms
rtt min/avg/max/mdev = 39.770/40.047/40.421/0.274 ms
2Kbyte packets:
rtr:~# ping -c 4 -s 2048 jolt.co.uk
PING jolt.co.uk (84.234.17.86) 2048(2076) bytes of data.
2056 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=1 ttl=125 time=47.6 ms
2056 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=2 ttl=125 time=46.7 ms
2056 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=3 ttl=125 time=46.9 ms
2056 bytes from bruce.ferrago.net (84.234.17.86): icmp_seq=4 ttl=125 time=47.0 ms
--- jolt.co.uk ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3003ms
rtt min/avg/max/mdev = 46.705/47.083/47.609/0.393 ms
I would expect gaming mode to have the same affect on the above tests (taking 8+ms off in my case).
Edited by deleted (Sun 22-Mar-09 10:25:54)