Re: Different Implementations of VLIW .
- From: eugene@xxxxxxxxxxxx (Eugene Miya)
- Date: 12 May 2008 14:35:10 -0800
In article <482518BB.B96B853C@xxxxxxxxxxxxxxxxxx>,
Eric P. <eric_pattison@xxxxxxxxxxxxxxxxxx> wrote:
Robert Myers wrote:wikipedia
On May 9, 6:53 pm, "Eric P." <eric_patti...@xxxxxxxxxxxxxxxxxx> wrote:
According to wiki, Multiflow started in 1984 and sold only
125 cpus before closing its doors 1990.
The moral seems mostly about the business prospects
for niche architectures.
Multiflows were not regarded as niche machines. They were regarded as
general purpose minisupecomputers.
Ditto.The commercial success or failure of a particular idea doesn't tell
you much of anything other than what a particular company did with a
particular idea at a particular time. No one here buys the idea that
alpha was a lousy architecture because it isn't available any more.
In general I appreciate and agree with your sentiments.
And if they had failed in the first year due to lack of funding,
or if they got orders for 250,000 units, expanded too quickly,
took on too much debt and then suffered a sudden downturn
in the economy, well c'est la vie.
But if you can only sell 125 units over a 6 year period,
it's because no one wants the damn things.
1) I signed a non-disclosure agreement, FWIW, with Multiflow.
2) The 1980s were a vastly different time and computer market then than now.
3) 125 units for that era wasn't a bad number. Frankly for the CHM, I
would love to have the list of 125 sites.
4) For context, of all those computers of that class and type Convex
came out ahead (I have a head count, but that's now HP proprietary).
They have their own dead news group. Alliant, ditto,
but a weaker architecture hardware and software. Lots of 80s firms had
machines with far fewer than 125 units: SAXPY (no wikipedia page),
ES-CD, possibly Flex (if one regards simple bus connected micros as
minisupers [marginal at best]). Oh and that other boat anchor KSR.
But I do think that more hypercubes of either Intel or NCUBE variety were
sold than Traces. None of those machines had audiences willing to push
a comp.sys.name-your-arch.
It was a candy store period of experimentation. It wasn't an easy time
to do business, but there was certainly venture capital.
People were interested in cost-performance, but a lot of people were
still fighting internal OS wars (a lot of firms didn't want to switch
from their vendor provided OSes). It was not as simple as not wanting
things.
--
.
- References:
- Re: Different Implementations of VLIW .
- From: Robert Myers
- Re: Different Implementations of VLIW .
- From: Eric P.
- Re: Different Implementations of VLIW .
- Prev by Date: Re: Different Implementations of VLIW .
- Next by Date: Re: a68 again was: VLIW .
- Previous by thread: Re: Different Implementations of VLIW .
- Next by thread: Re: Different Implementations of VLIW .
- Index(es):
Relevant Pages
|