Discussion:
Intel Itanium Processor (9700 series a.k.a. Kittson)
(too old to reply)
Neil Rieck
2017-05-11 15:56:57 UTC
Permalink
Raw Message
Intel Itanium Processor (9700 series a.k.a. Kittson)
http://ark.intel.com/products/family/451/Intel-Itanium-Processor

Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/dark_age.html
IanD
2017-05-11 16:50:13 UTC
Permalink
Raw Message
Post by Neil Rieck
Intel Itanium Processor (9700 series a.k.a. Kittson)
http://ark.intel.com/products/family/451/Intel-Itanium-Processor
Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/dark_age.html
What's it got over the previous cpu's?

Larger cache, and only a slightly higher clock speed. Anything else of significance?

The sooner VMS moves off Itanium the better. It is a shame the Itanium was never developed further but one could say the same for Alpha

Even AMD's up and coming Naples looks like it could give the Itanium a thumping. 32 cores (64 simultaneous threads), 8 memory channel with 2TB ram support, 128 PCIe 3.0 lanes! Let's hope the rewrite of VMS can take advantage of this cpu because its about to pop out of the AMD woodshed very soon

https://arstechnica.com/information-technology/2017/03/amd-naples-server-processor-more-cores-bandwidth-memory-than-intel/

Even the Ryzen 7 PC cpu has got 8 cores and comes in a 3.6 GHz (4.0 GHz Turbo) version, sure, it will not have the same I/O throughput as the Itanium but in other workloads I'm betting it will walk all over the Itanium

Are there any benchmarks that show the expected performance improvements of the new Itanium? The side by side comparison to the Paulson doesn't indicate much of an improvement at first glance
David Froble
2017-05-11 19:12:37 UTC
Permalink
Raw Message
Post by IanD
Post by Neil Rieck
Intel Itanium Processor (9700 series a.k.a. Kittson)
http://ark.intel.com/products/family/451/Intel-Itanium-Processor
Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/dark_age.html
What's it got over the previous cpu's?
Larger cache, and only a slightly higher clock speed. Anything else of significance?
The sooner VMS moves off Itanium the better. It is a shame the Itanium was never developed further but one could say the same for Alpha
Even AMD's up and coming Naples looks like it could give the Itanium a thumping. 32 cores (64 simultaneous threads), 8 memory channel with 2TB ram support, 128 PCIe 3.0 lanes! Let's hope the rewrite of VMS can take advantage of this cpu because its about to pop out of the AMD woodshed very soon
https://arstechnica.com/information-technology/2017/03/amd-naples-server-processor-more-cores-bandwidth-memory-than-intel/
Even the Ryzen 7 PC cpu has got 8 cores and comes in a 3.6 GHz (4.0 GHz Turbo) version, sure, it will not have the same I/O throughput as the Itanium but in other workloads I'm betting it will walk all over the Itanium
Are there any benchmarks that show the expected performance improvements of the new Itanium? The side by side comparison to the Paulson doesn't indicate much of an improvement at first glance
Understand, I'm not a fan of the itanic.

One of our customers has an i4, and routinely 500+ users. It's doing the job.

Now, for some, maybe not, but to me 500 users is a bit impressive.
abrsvc
2017-05-11 19:31:13 UTC
Permalink
Raw Message
Not to burst your bubble but...

Be careful just counting users. I used to easily support 45 users comfortably on a VAX 780. The I4 is much more powerful than 12 times a 780...

Any comparison has to include some description of what (if anything) the users are actually doing.

Dan
David Froble
2017-05-11 22:48:32 UTC
Permalink
Raw Message
Post by abrsvc
Not to burst your bubble but...
Be careful just counting users. I used to easily support 45 users comfortably on a VAX 780. The I4 is much more powerful than 12 times a 780...
Any comparison has to include some description of what (if anything) the users are actually doing.
Dan
No bubble to burst ..

Interactive users, and yeah, such can be on coffee break ..

Services accepting incoming connection requests and providing inquiry, order
processing, and such.

Background stuff.

Not something I plan on studying, just mentioning what's being supported.

But overall, it's rather busy.
Norman F Raphael
2017-05-12 02:30:13 UTC
Permalink
Raw Message
Sent: Thu, May 11, 2017 3:40 pm
Subject: Re: [New Info-vax] Intel Itanium Processor (9700 series a.k.a. Kittson)
Not to burst your bubble but...
Be careful just counting users. I used to easily support 45 users comfortably on a VAX 780. The I4 is much
more powerful than 12 times a 780...
Any comparison has to include some description of what (if anything) the users are actually doing.
Dan
Yep.

I had 750's with 8 DZ11s each in them (so potentially 64 users at once when the recommended load was
12-16 users) and 2 DZ11's was the maximum "supported" configuration in the box - the others were on a wall.

Those users were trained to login, do a task, then logout, but still every character interrupted the CPU....


Norman F. Raphael
"Everything worthwhile eventually
degenerates into real work." -Murphy
IanD
2017-05-14 09:24:44 UTC
Permalink
Raw Message
On Friday, May 12, 2017 at 5:12:39 AM UTC+10, David Froble wrote:

<snip>
Post by David Froble
Understand, I'm not a fan of the itanic.
One of our customers has an i4, and routinely 500+ users. It's doing the job.
Now, for some, maybe not, but to me 500 users is a bit impressive.
It is. I remember when I started off as an operator on VMS, one was lucky to see 30 users on a Vax!

On our system now we typically have around 900. Around Christmas that ramped up to just over 1K. That's on a i4 with a bucket load of ram

They are interactive sessions which often spin off batch jobs to complete workloads. The batch jobs are the system killers

The i2 before it simply couldn't keep up. The i4 does the job except when locking gets excessive, mysync ramps up. One cpu is dedicated to locking and it helps. I think if we were to add another 200 odd users the system would succumb to the drastic mysync death spiral

Better inter process syncing mechanisms are needed but that's not just VMS that suffers this. I was reading a users experience with windows 10 the other day with a trail of one of the AMD recently released cpu's and they claimed all the cpu's were showing 25% usage with little workload

I wonder what issues will come out of the woodwork when we go to x86 based cpu's with their higher clock speeds. Are there any race conditions awaiting in VMS I wonder
r***@gmail.com
2017-05-15 04:58:21 UTC
Permalink
Raw Message
Post by IanD
It is. I remember when I started off as an operator on VMS, one was lucky
to see 30 users on a Vax!
Utah State University had up to 64 students on a VAX, before they started
buying more VAXen. It would take an hour or two to log in, but once you
were in it would do OK.
--
roger ivie
***@gmail.com
Stephen Hoffman
2017-05-15 15:45:39 UTC
Permalink
Raw Message
Post by IanD
They are interactive sessions which often spin off batch jobs to
complete workloads. The batch jobs are the system killers
There's a long history of that happening with batch jobs. Memory
usage, locking contention, etc. Priority inversions does tend to
degrade interactive performance, though slogging through lock priority
problems with priority boosts does usually avoid the deadlocks.
Post by IanD
The i2 before it simply couldn't keep up. The i4 does the job except
when locking gets excessive, mysync ramps up. One cpu is dedicated to
locking and it helps. I think if we were to add another 200 odd users
the system would succumb to the drastic mysync death spiral
Synchronization issues are why gonzo numbers of cores aren't viable
designs for many applications, and why adding cores and sockets to a
server configuration can reduce performance. Cache-coherent
multiprocessors get slower as more cores are added, and designs such as
NUMA aren't a panacea.
Post by IanD
Better inter process syncing mechanisms are needed but that's not just
VMS that suffers this.
There's a decent range of those on OpenVMS, ranging from spinlocks to
message-passing to lock management to semaphores to turd files to...
Always better to have and to need less synchronization, where that's
possible. Sharding and partitioning of loads, where that's workable.
Scheduling batch off on cloned read-only data where that's feasible,
too. Distributed scheduling and process management are also weak
areas of OpenVMS, requiring local tools or third-party tools or the
fragility and complexity of scheduling via the queue manager. Having
something akin to libdispatch would be handy, as POSIX is a bit of a
slog — though it's certainly feasible to use the existing queue
primitives to get something akin to that, less the integration and
tooling.
Post by IanD
I was reading a users experience with windows 10 the other day with a
trail of one of the AMD recently released cpu's and they claimed all
the cpu's were showing 25% usage with little workload
Across various platforms, there are interesting discussions of what
measurements are taken and what particular measurements are showing,
too. Whether the systems are doing useful work, or are waiting in user
mode for something to happen, both can legitimately be considered user
activity, but the latter case masks what can be idle processors; where
some other aspect of the system configuration is the constraint and not
the processor. I look at what's available to developers and end-users
with Instruments and dtrace on macOS, and no easy analogs for most of
those tools over in OpenVMS.
Post by IanD
I wonder what issues will come out of the woodwork when we go to x86
based cpu's with their higher clock speeds. Are there any race
conditions awaiting in VMS I wonder
Faster processors have broken timing assumptions in VMS application
code since the second VAX, certainly. Or what issues around memory
page protections might arise, an area that VSI is undoubtedly exploring
as part of the port. Or what other mechanisms might become available
for coordinating activity or improving performance, for that matter.
--
Pure Personal Opinion | HoffmanLabs LLC
Bob Gezelter
2017-05-15 16:29:48 UTC
Permalink
Raw Message
Post by IanD
<snip>
Post by David Froble
Understand, I'm not a fan of the itanic.
One of our customers has an i4, and routinely 500+ users. It's doing the job.
Now, for some, maybe not, but to me 500 users is a bit impressive.
It is. I remember when I started off as an operator on VMS, one was lucky to see 30 users on a Vax!
On our system now we typically have around 900. Around Christmas that ramped up to just over 1K. That's on a i4 with a bucket load of ram
They are interactive sessions which often spin off batch jobs to complete workloads. The batch jobs are the system killers
The i2 before it simply couldn't keep up. The i4 does the job except when locking gets excessive, mysync ramps up. One cpu is dedicated to locking and it helps. I think if we were to add another 200 odd users the system would succumb to the drastic mysync death spiral
Better inter process syncing mechanisms are needed but that's not just VMS that suffers this. I was reading a users experience with windows 10 the other day with a trail of one of the AMD recently released cpu's and they claimed all the cpu's were showing 25% usage with little workload
I wonder what issues will come out of the woodwork when we go to x86 based cpu's with their higher clock speeds. Are there any race conditions awaiting in VMS I wonder
IanD,

If locking is high, the better question is "Locking WHAT?"

In addition to speed, volume/user scale up often uncovers issues which were not a consideration in smaller/lower volume configurations.

Metrics are only part of the answer in these cases. Often, only metrics in combination with a review of the codebase will uncover the underlying nature of the problem.

Maxing out a resource (e.g., memory, CPU, locking) is only a symptom. Increasing the resources is one option, but it often can be purely palliative (e.g., aspirin will reduce a fever; it does nothing for the underlying infection). I am reminded of a client site years ago (when memory was expensive). They had horrible response time, and the vendor recommendation was an expensive memory expansion.

The problem was that they had tens of users each running a multimegabyte applications image. The system was swapping like mad. The solution was a single command, INSTALL the applications image (it was written in FORTRAN and compiled as part of the installation procedures, so I had the source). This reduced the memory requirements from megabytes/user to a few hundred kilobytes/user. Swapping was eliminated, response time went back to reasonable.

- Bob Gezelter, http://www.rlgsc.com
Arne Vajhøj
2017-05-15 23:45:42 UTC
Permalink
Raw Message
Post by IanD
Post by David Froble
Understand, I'm not a fan of the itanic.
One of our customers has an i4, and routinely 500+ users. It's doing the job.
Now, for some, maybe not, but to me 500 users is a bit impressive.
It is. I remember when I started off as an operator on VMS, one was
lucky to see 30 users on a Vax!
On our system now we typically have around 900. Around Christmas that
ramped up to just over 1K. That's on a i4 with a bucket load of ram
I have seen more than 30 users on a VAX.

About 90 on VAX 8650 and about 110 on a VAX 6420.

Mostly editing, building and testing code.

Late 80's.

Arne

Kerry Main
2017-05-12 00:56:11 UTC
Permalink
Raw Message
-----Original Message-----
via Info-vax
Sent: May 11, 2017 12:50 PM
Subject: Re: [Info-vax] Intel Itanium Processor (9700 series a.k.a.
Kittson)
Post by Neil Rieck
Intel Itanium Processor (9700 series a.k.a. Kittson)
http://ark.intel.com/products/family/451/Intel-Itanium-Processor
Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/dark_age.html
What's it got over the previous cpu's?
Larger cache, and only a slightly higher clock speed. Anything else of significance?
The sooner VMS moves off Itanium the better. It is a shame the Itanium
was never developed further but one could say the same for Alpha
Even AMD's up and coming Naples looks like it could give the Itanium a
thumping. 32 cores (64 simultaneous threads), 8 memory channel with
2TB ram support, 128 PCIe 3.0 lanes! Let's hope the rewrite of VMS
can
take advantage of this cpu because its about to pop out of the AMD
woodshed very soon
https://arstechnica.com/information-technology/2017/03/amd-naples-
server-processor-more-cores-bandwidth-memory-than-intel/
Even the Ryzen 7 PC cpu has got 8 cores and comes in a 3.6 GHz (4.0
GHz
Turbo) version, sure, it will not have the same I/O throughput as the
Itanium but in other workloads I'm betting it will walk all over the
Itanium
Are there any benchmarks that show the expected performance
improvements of the new Itanium? The side by side comparison to the
Paulson doesn't indicate much of an improvement at first glance
While everyone likes to talk about server performance, the reality is
that 75% of most applications today could likely be handled with just
about any current server HW platform today.

Certainly 75+% of most OpenVMS Customers today are likely more than
happy with current I4/I2 systems performance.

Heck, Alpha system Customers typically are upgrading for financial
support/license/future protection reasons, not performance.

Having stated this, the new I6 systems will be of interest to the small
number of Customers that do need the likely 10-20% performance increase.

The availability of I6 systems is much more of interest to performance
challenged HP-UX Customers because HP-UX really only scales up, not out.

Regards,

Kerry Main
Kerry dot main at starkgaming dot com
Neil Rieck
2017-05-13 01:20:41 UTC
Permalink
Raw Message
Post by IanD
Post by Neil Rieck
Intel Itanium Processor (9700 series a.k.a. Kittson)
http://ark.intel.com/products/family/451/Intel-Itanium-Processor
Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/dark_age.html
What's it got over the previous cpu's?
Larger cache, and only a slightly higher clock speed. Anything else of significance?
The sooner VMS moves off Itanium the better. It is a shame the Itanium was never developed further but one could say the same for Alpha
Even AMD's up and coming Naples looks like it could give the Itanium a thumping. 32 cores (64 simultaneous threads), 8 memory channel with 2TB ram support, 128 PCIe 3.0 lanes! Let's hope the rewrite of VMS can take advantage of this cpu because its about to pop out of the AMD woodshed very soon
https://arstechnica.com/information-technology/2017/03/amd-naples-server-processor-more-cores-bandwidth-memory-than-intel/
Even the Ryzen 7 PC cpu has got 8 cores and comes in a 3.6 GHz (4.0 GHz Turbo) version, sure, it will not have the same I/O throughput as the Itanium but in other workloads I'm betting it will walk all over the Itanium
Are there any benchmarks that show the expected performance improvements of the new Itanium? The side by side comparison to the Paulson doesn't indicate much of an improvement at first glance
Not sure I agree. Itanium2 is currently the only game in town for OpenVMS as the x86-64 flavor is not yet born. Although Alpha served us well, our DS20e (which we purchased in 2002) was showing signs of age so moving to an rx2800-i2 last year bought us some insurance. It came with three years of support (hardware and software) from HPE which means we should be good until next year when we will swing our OpenVMS support contract over to VSI. Then on to x86-64 6-12 months after VSI releases it.

Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/technological_change.html
David Froble
2017-05-13 02:33:32 UTC
Permalink
Raw Message
Post by Neil Rieck
Not sure I agree. Itanium2 is currently the only game in town for OpenVMS as
the x86-64 flavor is not yet born. Although Alpha served us well, our DS20e
(which we purchased in 2002) was showing signs of age so moving to an
rx2800-i2 last year bought us some insurance. It came with three years of
support (hardware and software) from HPE which means we should be good until
next year when we will swing our OpenVMS support contract over to VSI. Then
on to x86-64 6-12 months after VSI releases it.
Neil Rieck Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/technological_change.html
Not sure what you're saying Neil? Not sure what an Itanium2 is?

Now, you've got an i2, and yes, they worked well for us. But we find the i4 to
be a bit better, nothing to get terribly excited about, but a definite
improvement. Note, if you're talking HPE and VMS, then yes, the i2 is as far as
they went. But they are re-selling the VSI products that support the i4, and
will be able to re-sell the VSI products for the i6. I'm guessing that if
someone is not getting what they need, the i6 may help a bit.
Camiel Vanderhoeven
2017-05-13 09:26:14 UTC
Permalink
Raw Message
Post by David Froble
Not sure what you're saying Neil? Not sure what an Itanium2 is?
Itanium 2 is intel's brand name for all Itanium processors after Merced.The decal attached to the front of some systems says "Intel Itanium 2 Inside".
David Froble
2017-05-13 14:01:19 UTC
Permalink
Raw Message
Post by Camiel Vanderhoeven
Post by David Froble
Not sure what you're saying Neil? Not sure what an Itanium2 is?
Itanium 2 is intel's brand name for all Itanium processors after Merced.The decal attached to the front of some systems says "Intel Itanium 2 Inside".
Ah, thank you. I don't have an itanic here, though the boss is threatening me
with one, so no actual physical hands on. Wasn't aware of the naming. "Itanic"
has always been good enough for me.

:-)
Neil Rieck
2017-05-13 12:26:14 UTC
Permalink
Raw Message
Post by David Froble
Not sure what you're saying Neil? Not sure what an Itanium2 is?
Now, you've got an i2, and yes, they worked well for us. But we find the i4 to
be a bit better, nothing to get terribly excited about, but a definite
improvement. Note, if you're talking HPE and VMS, then yes, the i2 is as far as
they went. But they are re-selling the VSI products that support the i4, and
will be able to re-sell the VSI products for the i6. I'm guessing that if
someone is not getting what they need, the i6 may help a bit.
Sir, you are correct. If it weren't for the efforts of VSI there would be no discussions anywhere about using OpenVMS on i4 (Itanium 9500 a.k.a. Poulson). For whatever reason, HP (now HPE) ended qualification tests of OpenVMS on Poulson signalling to everyone that they were going to let OpenVMS die. It was VSI who qualified OpenVMS on Poulson and have already announced their intent to support OpenVMS on Kittson.

[[[ Personal comment: I had been leading a 7-year effort to replace our Alphas with Itaniums in our shop. During that time, my employer did not know that HP had announced their intent to not support OpenVMS on Poulson (or anything after). They also did not know that VSI stepped into to save OpenVMS from extinction in 2014. If they had gotten wind of any of this then I am certain we would have been force to move off OpenVMS and possibly off HP hardware. So next year when our OpenVMS support contract with HPE ends, I'm going to have to sell it to them as "VSI is HPE's preferred vendor of software support" ]]]

Back to my original point: Itanium2 is the only game in town until VSI releases a version of OpenVMS that will run on x86-64.

Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/OpenVMS-Programmers-Corner.html
Dirk Munk
2017-05-13 09:43:05 UTC
Permalink
Raw Message
Post by IanD
Post by Neil Rieck
Intel Itanium Processor (9700 series a.k.a. Kittson)
http://ark.intel.com/products/family/451/Intel-Itanium-Processor
Neil Rieck
Waterloo, Ontario, Canada.
http://www3.sympatico.ca/n.rieck/docs/dark_age.html
What's it got over the previous cpu's?
Larger cache, and only a slightly higher clock speed. Anything else of significance?
The sooner VMS moves off Itanium the better. It is a shame the Itanium was never developed further but one could say the same for Alpha
Even AMD's up and coming Naples looks like it could give the Itanium a thumping. 32 cores (64 simultaneous threads),
Careful, there are some in this group who are adamant that a VMS system
will never need more than 2, or perhaps 4 cores and 640kB of memory. Or
am I mixing things up now?
Post by IanD
8 memory channel with 2TB ram support, 128 PCIe 3.0 lanes! Let's hope the rewrite of VMS can take advantage of this cpu because its about to pop out of the AMD woodshed very soon
https://arstechnica.com/information-technology/2017/03/amd-naples-server-processor-more-cores-bandwidth-memory-than-intel/
Even the Ryzen 7 PC cpu has got 8 cores and comes in a 3.6 GHz (4.0 GHz Turbo) version, sure, it will not have the same I/O throughput as the Itanium but in other workloads I'm betting it will walk all over the Itanium
Are there any benchmarks that show the expected performance improvements of the new Itanium? The side by side comparison to the Paulson doesn't indicate much of an improvement at first glance
Loading...