Despite big wins, AMD still not getting much respect on corporatedesktops

  • Thread starter Thread starter Yousuf Khan
  • Start date Start date
I found this interesting:

» Despite big wins, AMD still not getting much respect on corporate
desktops | Between the Lines | ZDNet.com
http://blogs.zdnet.com/BTL/index.php?p=1455

Interesting but not at all surprising. The main issue, as has so
often been the case, has little to do with the processors and
everything to do with the chipsets.

First problem: Until recently there were no AMD chipsets with
integrated graphics. While this might not seem like an issue for us
home users who tend to prefer external graphics, *ALL* corporate
desktop systems use them. Doing so not only reduces costs but it also
simplifies support of these systems, a key factor for corporate
desktops. If you look at the new Athlon64/Sempron based system that
HP offers in their corporate desktop line (the dx5150) they are based
on the relatively new ATI Radeon Xpress 200 chipset using ATI's
integrated graphics.

For those who haven't seen it, this HP dx5150 system is a pretty damn
kick-ass system for a corporate desktop.. better performance and
feature set (integrated dual-head video is rather nifty!) than the
much more expensive Intel-based dc7100 also mentioned this article.
The only real downside, and the reason (IMO) why it's classed as a
mid-range 5000 series instead of a high-end 7000 series business
desktop is that it uses a different BIOS than HP's standard. This
means that some of the nifty remote manageability tools might not work
right (I haven't tested this one way or the other, so I can't say for
sure what, if anything, breaks). It's also missing a few optional
security feature add-ons available for the 7000-series, ie the hood
lock and the TPM/TCPA module, though these are more design decisions
on HP's part rather than any failing of the processor or chipset (the
Intel-based 5000 series systems also lack these features).

Second problem: Different chipset vendors equals a whole can of worms
for corporate desktops. The article hits on this, and actually it's
one thing that won AMD some fans a while back. The H&R block case
they mentioned was specifically for the HP d315 and it's follow-up
d325 systems. These both used AthlonXP processors and nVidia chipset
(the nForce 220 in the case of the d315 and the nForce2 on the d325).
AMD and nVidia trumpeted the H&R Block as a real win for themselves
because the company was able to use an identical image on both of
these systems, something that is damn near impossible to do normally.
Obviously Intel learned from this and is working on stabilizing their
drivers sufficiently that they can do the same sort of thing.

Unfortunately though that just wasn't going to be an option going
forward, and switching your image from one Intel chipset to another
Intel chipset is easier than going from an nVidia chipset to an ATI
one. With the two Intel images you can usually use all of the same
drivers but just re-install them while building your image. With
nVidia vs. ATI you need a whole separate set of drivers from the
ground up.
 
Tony said:
Interesting but not at all surprising. The main issue, as has so
often been the case, has little to do with the processors and
everything to do with the chipsets.

The various manufacturers simply have to settle on a single supplier of
chipsets and/or motherboards in the case of AMD systems. For example,
Gateway may decide to go with SiS, HP with VIA, somebody else Nvidia,
somebody else ATI, etc. But they stick to that choice for a stable
period of time -- that way they can build in their various system
management tools around it. The abundance of choice in the AMD platform
shouldn't be used as a negative excuse to exclude it from use. Just
because Intel has restricted competition in its own platforms forcing
everyone to into a defacto single-choice corporate desktop, doesn't mean
you can't adhere to some single-standard in the AMD world -- you just
have to make your choice and stick to it.

Besides, I don't think it's really the chipsets that are the problem
here. It has more to do with Intel restricting competition in this
marketplace. Intel tends to target some markets as strategic to its
business image, and it restricts manufacturers from offering any
competing brands in that marketplace. It will tolerate competition in
the home market but not in the work market where its nameplate will be
highly visible.

Yousuf Khan
 
The various manufacturers simply have to settle on a single supplier of
chipsets and/or motherboards in the case of AMD systems. For example,
Gateway may decide to go with SiS, HP with VIA, somebody else Nvidia,
somebody else ATI, etc. But they stick to that choice for a stable
period of time -- that way they can build in their various system
management tools around it.

HP did just that with nVidia. The old d315 and d325 lines were rather
long-lived, stable designs. In fact, the d325 is still available
today and has been as long-lived and stable of a product line as any
of their Intel-based systems. The problem was that nVidia just didn't
have a viable follow-up chipset due to the lack of integrated
graphics. In fact, they still don't have such a chipset for the
Athlon64 and aren't expected to for a while yet.
The abundance of choice in the AMD platform
shouldn't be used as a negative excuse to exclude it from use. Just
because Intel has restricted competition in its own platforms forcing
everyone to into a defacto single-choice corporate desktop, doesn't mean
you can't adhere to some single-standard in the AMD world -- you just
have to make your choice and stick to it.

I think it's more a question of Intel having a wider range of chipset
options to fill all the needed markets. nVidia really only 3
configurations for desktop chipsets for the Athlon64. Intel currently
has about 15 versions of their various current chipsets for P4 chips,
many of which can be used fairly interchangeably from a design
standpoint but allowing companies to hit different markets and
price-points.
Besides, I don't think it's really the chipsets that are the problem
here. It has more to do with Intel restricting competition in this
marketplace. Intel tends to target some markets as strategic to its
business image, and it restricts manufacturers from offering any
competing brands in that marketplace. It will tolerate competition in
the home market but not in the work market where its nameplate will be
highly visible.

There may be some backroom dealings involved, I have absolutely no
connection to those sorts of decisions so I really don't know. There
are definitely also some marketing issues, I STILL meet lots of people
in the business community that just don't trust AMD processors.
Whether it be that they fear some sort of odd-ball incompatibility
(I've found the same number of incompatibilities with Intel chips as
with AMD chips) or they feel that the processors have a much higher
failure rate (in my experience both Intel and AMD chips have VERY low
failure rates).

However there definitely have been and continue to be some technical
issues that have made AMD chips less attractive in certain markets. I
can say with a high degree of certainty that the lack of good
integrated graphics chipset for the Athlon64 hurt it's potential in
the corporate market. It'll be interesting to see if HP brings out a
new nVidia/AMD based system when nVidia releases their (just announced
this week) integrated graphics Athlon64 chipset later this year.
 
Tony said:
HP did just that with nVidia. The old d315 and d325 lines were rather
long-lived, stable designs. In fact, the d325 is still available
today and has been as long-lived and stable of a product line as any
of their Intel-based systems. The problem was that nVidia just didn't
have a viable follow-up chipset due to the lack of integrated
graphics. In fact, they still don't have such a chipset for the
Athlon64 and aren't expected to for a while yet.

It's surprising that Nvidia still hasn't found a suitable successor to
its Nforce2 chipset. All subsequent generations were less capable than
the Nforce2.

There may be some backroom dealings involved, I have absolutely no
connection to those sorts of decisions so I really don't know. There
are definitely also some marketing issues, I STILL meet lots of people
in the business community that just don't trust AMD processors.

That's entirely due to marketing. If it's on tv, then it must be good.
IT people are just as likely to succumb to that as managers.

However there definitely have been and continue to be some technical
issues that have made AMD chips less attractive in certain markets. I
can say with a high degree of certainty that the lack of good
integrated graphics chipset for the Athlon64 hurt it's potential in
the corporate market. It'll be interesting to see if HP brings out a
new nVidia/AMD based system when nVidia releases their (just announced
this week) integrated graphics Athlon64 chipset later this year.

I doubt if they even need "good" integrated graphics, just integrated
graphics in general would suffice.

Yousuf Khan
 
There may be some backroom dealings involved, I have absolutely no
connection to those sorts of decisions so I really don't know. There
are definitely also some marketing issues, I STILL meet lots of people
in the business community that just don't trust AMD processors.

As Yousuf and George are well aware, that persistent marketplace
irrationality would disappear if only I would stop my cranky posts
insisting that buyers prefer market leaders. If Intel doesn't soon
start paying me for my slanted posts, I will stop making them, and the
market for Intel processors will collapse.

I'd be impressed if Intel's success at imposing market discipline were
nearly as effective as Yousuf seems to think it is. Standard Oil did
it successfully and got busted, IBM did it successfully and got
busted, Microsoft did it successfully and got busted (but what a joke
that was). I guess that's why Yousuf is jumping up and down with
excitement at the enforcement action in Japan. If Intel's success is
due to anticompetitive practices, it will get busted. I'm skeptical,
to say the least. Actually, I think it's clear I believe Yousuf is
just plain wrong, but I always allow for the possibility that I could
be wrong.
Whether it be that they fear some sort of odd-ball incompatibility
(I've found the same number of incompatibilities with Intel chips as
with AMD chips) or they feel that the processors have a much higher
failure rate (in my experience both Intel and AMD chips have VERY low
failure rates).

However there definitely have been and continue to be some technical
issues that have made AMD chips less attractive in certain markets. I
can say with a high degree of certainty that the lack of good
integrated graphics chipset for the Athlon64 hurt it's potential in
the corporate market. It'll be interesting to see if HP brings out a
new nVidia/AMD based system when nVidia releases their (just announced
this week) integrated graphics Athlon64 chipset later this year.
Intel's larger volume means that it can afford more development effort
for things like chipsets. That's just one of the advantages the
market leader has automatically.

RM
 
YKhan said:
It's surprising that Nvidia still hasn't found a suitable successor to
its Nforce2 chipset. All subsequent generations were less capable than
the Nforce2.

NVidia never captured the market share they expected with the IGP
versions of nForce and nForce2 and they explicitly stated that as
the reason for dropping the IGP from nForce3.
That's entirely due to marketing. If it's on tv, then it must be good.
IT people are just as likely to succumb to that as managers.

Most IT people I know are too smart for that. They do, however,
still have to buy Intel a lot simply because the PHBs they report
to are morons.

For example, a friend of mine recently started changing his
servers from Opterons to Xeons simply because his new boss liked
the commercial with "the blue guys in the climbing gear". It
didn't matter that they were going to need 40% more servers and
twice the air conditioning gear if they switched to Xeons: idiot
bosses make idiotic decisions.

Another friend, however, is thrilled to be buying those 10 month
old used Opteron servers for the company he works for. His boss
is just as ignorant, IT-wise, but at least she is smart enough to
know she is ignorant and lets the trained and experienced pros
make the decisions that she doesn't have the background for.
 
Rob said:
Another friend, however, is thrilled to be buying those 10 month old
used Opteron servers for the company he works for. His boss is just as
ignorant, IT-wise, but at least she is smart enough to know she is
ignorant and lets the trained and experienced pros make the decisions
that she doesn't have the background for.

Actually, to me that would be the definition of a smart manager.
Somebody who recognizes their ignorance.

Yousuf Khan
 
Robert said:
I'd be impressed if Intel's success at imposing market discipline were
nearly as effective as Yousuf seems to think it is. Standard Oil did
it successfully and got busted, IBM did it successfully and got
busted, Microsoft did it successfully and got busted (but what a joke
that was). I guess that's why Yousuf is jumping up and down with
excitement at the enforcement action in Japan. If Intel's success is
due to anticompetitive practices, it will get busted. I'm skeptical,
to say the least. Actually, I think it's clear I believe Yousuf is
just plain wrong, but I always allow for the possibility that I could
be wrong.

You may think as you like. The Japanese action was the first concrete
proof of something that's been evident for a long long time. Actually,
it wasn't just "proof" it was a confirmation, because Intel admitted to
it over there.

Yousuf Khan
 
As Yousuf and George are well aware, that persistent marketplace
irrationality would disappear if only I would stop my cranky posts
insisting that buyers prefer market leaders. If Intel doesn't soon
start paying me for my slanted posts, I will stop making them, and the
market for Intel processors will collapse.

Nobody has said there has to be a "collapse" - that's just your active
imagination conjuring up delusions again. If the market preferred the
highest volume vendor, we'd only have one car company, one of everything in
fact.
I'd be impressed if Intel's success at imposing market discipline were
nearly as effective as Yousuf seems to think it is. Standard Oil did
it successfully and got busted, IBM did it successfully and got
busted, Microsoft did it successfully and got busted (but what a joke
that was). I guess that's why Yousuf is jumping up and down with
excitement at the enforcement action in Japan. If Intel's success is
due to anticompetitive practices, it will get busted. I'm skeptical,
to say the least. Actually, I think it's clear I believe Yousuf is
just plain wrong, but I always allow for the possibility that I could
be wrong.

We live in different times, which is why M$ only got their ear twisted;
hell even IBM got only a slap on the wrist and got rid of one of their
pending liabilities into the bargain, as part of the settlement. Now you
expect us to believe that what you think trumps what the Japanese FTC has
concluded... not that I expect the US FTC to reach the same conclusion.
Now that the man who said "They buy the Megahertz" has gone, even Intel has
confessed that they are beat technically.
Intel's larger volume means that it can afford more development effort
for things like chipsets. That's just one of the advantages the
market leader has automatically.

I don't see the chipsets thing - Intel only has a couple of chipsets per
sector with variations like IGP. In fact it's only recently that they had
a server chipset worth anything - it's also the chipsets architecture which
is the reason they're getting beaten in performance. Now they've welcomed
nVidia to their "platform" gambit since they can't do that bit themselves -
doomed to more failure IMO... nVidia will be gone when they realize they're
only allowed breadcrumbs for lunch. According to the proposed plot so far,
they won't get a look at the business sector anyway, since they'll be a
"different image".Ô_ô

The success stories are starting to roll in Robert: The Weather Channel,
Bell Helicopter, Verisign and of course M$ themselves love Opteron with the
huge successes in the MSN server farm and MS Treasury. When the principle
OS vendor endorses the hardware, especially for big-time server ops, that
really counts. What some Weary Willie in IT or management thinks right now
doesn't matter - those people are malleable or replaceable.
 
Nobody has said there has to be a "collapse" - that's just your active
imagination conjuring up delusions again. If the market preferred the
highest volume vendor, we'd only have one car company, one of everything in
fact.
Markets do change. For all I know, Intel could be in a serious
downward spiral. I don't think that I, or anyone else, for that
matter, is bright enough to see that with any real clarity or
certainty, whether from inside or outside Intel.

Markets do prefer the highest volume vendor, but to go from that to
saying that there would be just one vendor as a result just doesn't
follow. You know that, so why are you presenting such an absurdity as
an argument?
We live in different times, which is why M$ only got their ear twisted;
hell even IBM got only a slap on the wrist and got rid of one of their
pending liabilities into the bargain, as part of the settlement. Now you
expect us to believe that what you think trumps what the Japanese FTC has
concluded... not that I expect the US FTC to reach the same conclusion.

I'm not even sure what it is you think I believe that I expect you to
believe. Intel pushed the envelope in Japan? I don't doubt it. How
serious a legal problem is it for them? I really don't have a way to
evaluate. I haven't seen any evidence that it is a serious legal
problem for them. What is it, exactly, that I expect you to believe?
Now that the man who said "They buy the Megahertz" has gone, even Intel has
confessed that they are beat technically.
For the moment.
I don't see the chipsets thing - Intel only has a couple of chipsets per
sector with variations like IGP. In fact it's only recently that they had
a server chipset worth anything - it's also the chipsets architecture which
is the reason they're getting beaten in performance.

You mean because the memory controller isn't on board or because of
hypertransport, which isn't a memory interface (joke--for the other
high-stepper here)? I don't understand why Intel has been so slow to
move on connectivity like hypertransport. The decision to keep the
memory controller off the chip is clearly a business decision, not a
technical decision.
Now they've welcomed
nVidia to their "platform" gambit since they can't do that bit themselves -
doomed to more failure IMO... nVidia will be gone when they realize they're
only allowed breadcrumbs for lunch. According to the proposed plot so far,
they won't get a look at the business sector anyway, since they'll be a
"different image".Ô_ô

The success stories are starting to roll in Robert: The Weather Channel,
Bell Helicopter, Verisign and of course M$ themselves love Opteron with the
huge successes in the MSN server farm and MS Treasury. When the principle
OS vendor endorses the hardware, especially for big-time server ops, that
really counts. What some Weary Willie in IT or management thinks right now
doesn't matter - those people are malleable or replaceable.

Hadn't heard about the MSN server farm. I've been pondering the
Apple/Intel deal. I guess we'll find out what that's really about,
but if Apple puts OS X on x86, that makes Apple a direct competitor
for Microsoft. There are only two reasons for Apple not to be an
aggressive competitor that I can think of: Microsoft Office and money.
I don't know what to say about Microsoft Office, but, with Intel on
the team, money shouldn't be a problem.

As to what those Wearie Willie or his ignorant boss thinks, markets
move slowly, but they do move. From outside DEC, it sure looked as if
DEC thought it was irreplaceable. DEC was wrong, of course, but the
chip business has as much to do with investment capital as it has to
do with technical excellence and it would take alot to push Intel
aside. In the meantime, life moves on.

RM
 
Robert said:
I'm not even sure what it is you think I believe that I expect you to
believe. Intel pushed the envelope in Japan? I don't doubt it. How
serious a legal problem is it for them? I really don't have a way to
evaluate. I haven't seen any evidence that it is a serious legal
problem for them. What is it, exactly, that I expect you to believe?

What? Since when does a legal judgement from a major country's
regulatory body not be a serious problem?

Well just to give you an example. They have now been prohibited from
entering into long-term contracts with OEMs, and they must now submit
all of their contracts to the FTC for scrutiny. I think this lasts for
two years, and can be extended beyond.
You mean because the memory controller isn't on board or because of
hypertransport, which isn't a memory interface (joke--for the other
high-stepper here)? I don't understand why Intel has been so slow to
move on connectivity like hypertransport. The decision to keep the
memory controller off the chip is clearly a business decision, not a
technical decision.

Hypertransport didn't happen overnight for AMD, so it's not going to
happen overnight for Intel either. Especially not the cache-coherent
extensions to HT.
Hadn't heard about the MSN server farm. I've been pondering the
Apple/Intel deal. I guess we'll find out what that's really about,
but if Apple puts OS X on x86, that makes Apple a direct competitor
for Microsoft. There are only two reasons for Apple not to be an
aggressive competitor that I can think of: Microsoft Office and money.
I don't know what to say about Microsoft Office, but, with Intel on
the team, money shouldn't be a problem.

Perhaps with virtualization, it's not going to matter as much if a
particular application doesn't exist on your own platform?

Yousuf Khan
 
Robert Myers wrote:

Hypertransport didn't happen overnight for AMD, so it's not going to
happen overnight for Intel either. Especially not the cache-coherent
extensions to HT.
I'd be astonished if such a thing weren't well down the pipeline at
Intel already. They're pretty good about keeping secrets.
Perhaps with virtualization, it's not going to matter as much if a
particular application doesn't exist on your own platform?
Don't know about virtualization (remember x86 and Itanium?). On the
other hand, I don't think Gates could pull Office away from OS X
without some kind of regulatory retaliation. It would be a great
scenario: one way or the other Gates-and-Ballmer-We-Are-the-World
would lose.

RM
 
Robert said:
I'd be astonished if such a thing weren't well down the pipeline at
Intel already. They're pretty good about keeping secrets.

Oh, they are working on it alright, it's called CSI. It's still not
going to happen overnight.

They've already lost considerable time by diverting and inventing
PCI-Express. Remember how in the early days that Intel was warning
people away from Hypertransport by saying that soon PCIe would be here
and much faster, better than HT? Well after all of that effort, PCIe
cannot be used as a processor-to-processor interconnect, it's too
complex. They couldn't get the main PCIe standards to become an
interconnect, so they decided to split the standard off and make the
more complex processor-to-processor stuff become the ASI standard
(Advanced Switching Interconnect) extension to PCIe. Now it's looking
like ASI is too complex too, so they're now going back to the drawing
board and coming up with a simpler interconnect, much more similar to
Hypertransport, called CSI.
Don't know about virtualization (remember x86 and Itanium?). On the
other hand, I don't think Gates could pull Office away from OS X
without some kind of regulatory retaliation. It would be a great
scenario: one way or the other Gates-and-Ballmer-We-Are-the-World
would lose.

What I'm saying is that if Microsoft withdraws support for Office on
MacOS on PPC, then the new x86 Macintoshes can simply run the x86
Windows version of Office in a separate virtual partition.

Yousuf Khan
 
Markets do change. For all I know, Intel could be in a serious
downward spiral. I don't think that I, or anyone else, for that
matter, is bright enough to see that with any real clarity or
certainty, whether from inside or outside Intel.

Markets do prefer the highest volume vendor, but to go from that to
saying that there would be just one vendor as a result just doesn't
follow. You know that, so why are you presenting such an absurdity as
an argument?

Hadn't you noticed? The FTC is asleep at the wheel... catatonic would be a
better description. All around us, the large business structures are
consolidating: banks, petroleum, autos, etc. with barely a peep from the
regulators.
I'm not even sure what it is you think I believe that I expect you to
believe. Intel pushed the envelope in Japan? I don't doubt it. How
serious a legal problem is it for them? I really don't have a way to
evaluate. I haven't seen any evidence that it is a serious legal
problem for them. What is it, exactly, that I expect you to believe?

I think the Japanese FTC action is serious for Intel - they had better
behave or there will be disastrous consequences. Their smug
business-as-usual reaction is not a good sign for them - they're a company
which shows almost palpable signs of corpulent rot.
For the moment.


You mean because the memory controller isn't on board or because of
hypertransport, which isn't a memory interface (joke--for the other
high-stepper here)? I don't understand why Intel has been so slow to
move on connectivity like hypertransport. The decision to keep the
memory controller off the chip is clearly a business decision, not a
technical decision.

The on-board memory controller is a large part of the AMD performance
advantage - see Keith about latency.:-) Hypertransport is the one chink I
see in AMD's strategy: the roadmap shows it stuck at 1000MHz well into 2006
and no sign of a ramp up at all. I don't think they can afford that.
Hadn't heard about the MSN server farm. I've been pondering the
Apple/Intel deal. I guess we'll find out what that's really about,
but if Apple puts OS X on x86, that makes Apple a direct competitor
for Microsoft. There are only two reasons for Apple not to be an
aggressive competitor that I can think of: Microsoft Office and money.
I don't know what to say about Microsoft Office, but, with Intel on
the team, money shouldn't be a problem.

For the kind of money involved, I don't see that Intel would care much but
if Apple thinks they can do the transition, I'm going have to seriously
look at the correct point to dump some Apple stock I bought ~15 years
ago.:-)

What seems the least credible part of the Apple rumors is that they would
openly talk of phasing out the IBM Power chips; a much better tactic, if
Intel is to be a part of their strategy, would be to talk of parallel lines
of systems for the foreseeable future. In fact it could be that they are
looking at Intel as a CPU supplier for some new consumer appliance - they
*are* an appliance innovator more than anything else.
As to what those Wearie Willie or his ignorant boss thinks, markets
move slowly, but they do move. From outside DEC, it sure looked as if
DEC thought it was irreplaceable. DEC was wrong, of course, but the
chip business has as much to do with investment capital as it has to
do with technical excellence and it would take alot to push Intel
aside. In the meantime, life moves on.

DEC, who latterly hated that name -- "Digital please" -- got hurt badly
because Intel stole their shiny new high-clock rate CPU technology - add in
the marketing blunders with Alpha and the fact that their service org
thought they could compete with IBM. It also seemed to take a while for
them to realize VAX was obsolete - possibly an architecture overhaul there
late 80s would have helped?

They lost their competitiveness and there *were* strong challengers in the
market - in some ways, possibly Sun got the market Digital should have had
the vision to chase... ironic that one of the founders came from Data
General.:-) Intel is kinda different because they don't sell directly to
their customers but if there were, e.g., a strong adversary against Dell
with a full supply of AMD CPUs, Intel could have found itself in a heap of
trouble by now after its recent missteps. That day may yet still come...
or Otellini may show Barrett up as a fumbler??
 
YKhan said:
What I'm saying is that if Microsoft withdraws support for Office on
MacOS on PPC, then the new x86 Macintoshes can simply run the x86
Windows version of Office in a separate virtual partition.

Yeah, right. The user has already shelled out for whatever OS
Apple decides to put on the x86 Mac - even if the cost of the OS
is hidden in the purchase price of the PC - and now you think the
user is going to spend more for virtualization software (a la
VMWare) and still more for a copy of Windows ?

Forget it. With rare exceptions, users who want to run Windows
apps on an x86 machine are going to ignore Macs completely unless
the machine will natively run Windows so they can skip the Mac OS
completely.
 
That's entirely due to marketing. If it's on tv, then it must be good.
IT people are just as likely to succumb to that as managers.

Sure marketing is important, though for a lot of people it will take
more than a few TV commercials.
I doubt if they even need "good" integrated graphics, just integrated
graphics in general would suffice.

I guess I should have qualified that by "good" I mean "Won't cause the
PC to crash due to crappy drivers" and not "high performance".
Admittedly by this definition Intel has, at times, struggled with
producing good integrated graphics for the desktop. However when the
only option for integrated graphics on the Athlon64 was VIA, then I
think HPaq made a wise choice to avoid them for a while.
 
Back
Top