|
PureBytes Links
Trading Reference Links
|
Your counts compare rather closely to those of mine(25 to 25% daily) over
the last two months with traders with a wide variety of computers and feeds
and locations. Here is a rambling summary of my search for what causes the
tick loss and what has been done to increase the count. The bottleneck IS
in the BMI data manager and or the way the GS hooks into it. Fact is when
you switch data feeds from BMI to DTN the tick counts on TS2K then exceed
that of TS4 on BMI. Not only that, with two computers and monitors sitting
side by side, BMI(TS4) lags DTN(TS2ki) by 7 seconds to a minute or more all
of the time. I have never seen BMI(TS4) lead DTN(TS2ki). There are times
when both feeds stall, DTN followed a few seconds later by BMI which
indicates a feed problem ahead of both of them. Some have suggested
switching to W2000 with TS2ki, and some have suggested using the
TurboExpress high speed comport. The story is that W95 and W2000 have less
tick loss than W98 and the 16550 mobo serial ports have overruns. However,
neither of those solutions increased the tick count on TS2ki on BMI for me
suggesting that the BMI DM was the culprit. TS4 on BMI W95 consistently
comes within a few percent of the ticks compared to TS2ki on DTN on W98
using the TurboExpress 920 card with the 16750 UARTs with the bigger buffer.
Fact is if you are using tick bar charts, the system signals lag when tick
loss occurs, common sense. For TS4(BMI) you may increase the tick count
depending on the vintage of the computer. Slower processors & motherboards
may benefit by lowering the receiver FIFO on the comport. Changing hardware
acceleration may help. Making the machine a "network"
(SystemProperties/Performance/FileSystem/select network server). Making the
machine a dedicated trading machine with no other programs running may help.
Especially turn off anti virus programs, empty the Startup file and only
open TS4. Have plenty of memory. Select a fixed swap file size and move it
to the front of the disk. Defrag. Fact is BMI feed filters ticks and there
is a limit to how much can be done to catch all that are fed to you. Those
who I compared tick counts with that had more counts on TS2ki with DTN than
I did had faster computers, some had the Turbo card and others did not.
Some had dual processor machines with scuzzi drives, others had UDMA66's.
All had plenty of memory to reduce(eliminate) swap files. One surprizing
result was esignal on cable with a fast modem. That came in neck and neck
with DTN on the tick count and sometimes(rarely) exceeded DTN. So, in my
book the most significant increase in count comes from dumping the BMI data
manager. Another surprising, maybe not so surprising thing, is that TS4 on
W95 is being run alongside TS2ki(DTN) by a few traders with no plans to shut
it down for the present. PCQuote via cable and fast modem is a favored way
by some. PCQuote via dialup modem can have problems if the ISP is not one
of the select few that it works well with. Best to ask PCQuote for info on
that. Finally, an Omega tech, frankly said, TS2ki needs a "powerful"
computer and the BMI Data Manager symbol conversion process is what slows it
down. There is no interest at OR or DBC in engineering a solution for a 38k
satellite feed. There are no plans, there is no interest, they don't want
to talk about it, nada, nix, fine`.
BobR
----- Original Message -----
From: <Odd@xxxxxxxx>
To: <omega-list@xxxxxxxxxx>
Sent: Sunday, April 30, 2000 10:21 PM
Subject: WHY IS TS2k LOSING TICKS?
> I have heard from several traders that TS2k is losing ticks left and
> right. On a BMI feed, the ES contract typically shows 20-30,000 ticks
> per day on TS4, while TS2k shows only 15-19,000 ticks on the exact same
> data feed!
> That is a tick loss of 15 - 35% !
>
> I don't know if the problem is with the BMI Data Manager - or if it is
> with the GlobalServer. It seems that the higher the number of ticks a
> symbol has, the bigger the percentage tick loss in TS2k. For a symbol
> with fewer ticks (for instance SP), the percentage loss is not as bad,
> even though it still is serious.
>
> I will encourage people who has both TS2k and TS4 to compare the tick
> count and see if this is happening on ALL data feeds, or if it is only
> related to the BMI data feed.
>
> To me, of all the reported problems with TS2k, this must be one of the
> most damaging. What good does it do, if you have the BEST system in the
> world, but you cannot trust your DATA?
>
> Cheers,
> Odd.
>
>
|