VMware workstation machine

[carnage519]

I have come to the conclusion that I need at least a Quad Core setup for my home workstation.. I'm upgrading my 2000 Pro/Server certifications to 2K8 soon, and have been working with 2K8 server a lot at work..

Would I notice a noticable performance increase between a PII 940 @ 3.0 Ghz vs a I7 920 @ 2.66 Ghz running two or three Vmware servers running along with a possible Windows 7 client running.. I plan on going with at least 8 GB of memory or possibly 16GB.. Will be running Raid 1 also on some Raid Edition WD drives or a Sea gate ES.

 

[SolMiester]

If using VMware, it will be hardware compatibility issues, not cpu performance and both CPU have equivalent virtual instruction sets and are about the same performance wise.

 

[Zstream]

VMware behaves better on AMD servers, at least 8+ way and above.

http://www.zdnet.com.au/review...402,339293122-3,00.htm

http://it.anandtech.com/IT/showdoc.aspx?i=3484&p=9

 

[Viditor]

Originally posted by: Zstream
VMware behaves better on AMD servers, at least 8+ way and above.

http://www.zdnet.com.au/review...402,339293122-3,00.htm

http://it.anandtech.com/IT/showdoc.aspx?i=3484&p=9

The new Shanghai Opterons specifically are designed for improved virtualization. Remember that's not PhII...it's Shanghai Opterons.

"Thanks to the lower world switch times, higher clock speed, and larger cache, the new "Shanghai" Opteron 8384 improves the already impressive scores of the AMD "Barcelona" 8356 by almost 43%. The only Intel that comes somewhat close is the hex-core behemoth known as the Xeon X7460, which needs a lot more power"

(from the AT link that Zstream posted)

BTW, you'll notice in AT's study that the 24 core Intel system (4 x 6-core) is beaten in VMMark by the 16 core (4 x 4-core) Shanghai system...only by 6%, but still that's quite impressive.

 

[jones377]

Originally posted by: carnage519
I have come to the conclusion that I need at least a Quad Core setup for my home workstation.. I'm upgrading my 2000 Pro/Server certifications to 2K8 soon, and have been working with 2K8 server a lot at work..

Would I notice a noticable performance increase between a PII 940 @ 3.0 Ghz vs a I7 920 @ 2.66 Ghz running two or three Vmware servers running along with a possible Windows 7 client running.. I plan on going with at least 8 GB of memory or possibly 16GB.. Will be running Raid 1 also on some Raid Edition WD drives or a Sea gate ES.

If you really need 16GB RAM then your only 2 choices are PhII or C2Q (for single sockets) as I don't think there are any 4GB DDR3 UNREG out yet, or if there are they would be VERY expensive. And since you're using VMWare I'd go with the Phenom2.

 

[aigomorla]

isnt VM ware based on parallism where number of threads actually play role?

if thats the case your telling the guy to go with a 4 threaded processor over a 8 threaded one.

You guys remember the i7 with HT that actually works can simulate 8 simulatanous threads.

Also 6 DDR3 slots vs 4.

 

[i7guy1]

An i7 will give you twice the number of vcpus. You can ask the question of vmwares' workstation forum, I'm sure the developers will chime in with the nod toward the i7.

At any rate the Nehalem server cores should obliterate the Shanghai cores.

 

[Viditor]

Originally posted by: aigomorla
isnt VM ware based on parallism where number of threads actually play role?

Not a dominant role...
For example, as AT found out, the 24-core Xeon X7460 system is 6% slower than the 16-core Opteron 8384 Shanghai system in VMMark.

Originally posted by: i7guy1
At any rate the Nehalem server cores should obliterate the Shanghai cores

Why? Remember that the design of Shanghai is specifically geared towards virtualization.

BTW, welcome to the board...

 

[Zstream]

Originally posted by: aigomorla
isnt VM ware based on parallism where number of threads actually play role?

if thats the case your telling the guy to go with a 4 threaded processor over a 8 threaded one.

You guys remember the i7 with HT that actually works can simulate 8 simulatanous threads.

Also 6 DDR3 slots vs 4.

*facepalm* Do any of you do actual server work?

 

[aigomorla]

Originally posted by: Zstream

Originally posted by: aigomorla
isnt VM ware based on parallism where number of threads actually play role?

if thats the case your telling the guy to go with a 4 threaded processor over a 8 threaded one.

You guys remember the i7 with HT that actually works can simulate 8 simulatanous threads.

Also 6 DDR3 slots vs 4.

*facepalm* Do any of you do actual server work?

As in VMWare? no but i was interested in setting one up on my gainestown.

Which is why im asking.

I Always thought more cores or thread application was better.

And thanks for the face palm.

 

[Zstream]

Originally posted by: aigomorla

Originally posted by: Zstream

Originally posted by: aigomorla
isnt VM ware based on parallism where number of threads actually play role?

if thats the case your telling the guy to go with a 4 threaded processor over a 8 threaded one.

You guys remember the i7 with HT that actually works can simulate 8 simulatanous threads.

Also 6 DDR3 slots vs 4.

*facepalm* Do any of you do actual server work?

As in VMWare? no but i was interested in setting one up on my gainestown.

Which is why im asking.

I Always thought more cores or thread application was better.

And thanks for the face palm.

More cores is better but I7 HT is still the same as P4 HT with slight modifications. Any many cases HT performs worse then nonHT. Such as in exchange, VM, sql 2005 etc..

VMware and HT perform better with little CPU load. If the VMware is above 25% load HT needs to be turned off.

 

[i7guy1]

That may have been true with previous cores, but I'm not sure about this with Nehalem. I run about 10 heavy duty application servers a mix of vms under ESX and physical boxes with up to 8 physical cores and lots of memory. Our vendor recommends HT off, but we've benchmarked and kept HT on.

 

[Zstream]

Originally posted by: i7guy1
That may have been true with previous cores, but I'm not sure about this with Nehalem. I run about 10 heavy duty application servers a mix of vms under ESX and physical boxes with up to 8 physical cores and lots of memory. Our vendor recommends HT off, but we've benchmarked and kept HT on.

Just stop please.... It is a well documented peformance hit, Microsoft has it off for crying out loud.

 

[nyker96]

there's no tests on them for vmware. maybe ask the company tech support they might give better recommendations of optimum hardware for their own products.

 

[KentState]

Originally posted by: Zstream

Originally posted by: aigomorla

Originally posted by: Zstream

Originally posted by: aigomorla
isnt VM ware based on parallism where number of threads actually play role?

if thats the case your telling the guy to go with a 4 threaded processor over a 8 threaded one.

You guys remember the i7 with HT that actually works can simulate 8 simulatanous threads.

Also 6 DDR3 slots vs 4.

*facepalm* Do any of you do actual server work?

As in VMWare? no but i was interested in setting one up on my gainestown.

Which is why im asking.

I Always thought more cores or thread application was better.

And thanks for the face palm.

More cores is better but I7 HT is still the same as P4 HT with slight modifications. Any many cases HT performs worse then nonHT. Such as in exchange, VM, sql 2005 etc..

VMware and HT perform better with little CPU load. If the VMware is above 25% load HT needs to be turned off.

Mr. Facepalm, I like how you state as a matter of fact that SQL Server performs worse under HT, when that is not necessarily the case. This should be tested by the organization for their particular instance. I've seen a substantial gain in performance using it with a data warehouse, various applications, and with a B2C website. Even MS states that a 10-20% improvement can be gained from using it.

http://support.microsoft.com/kb/322385

 

[i7guy1]

Different strokes for different applications. Since you don't know what I manage, and I'm not saying, you can either take my comments at face value, or not. Either way it doesn't matter.

I have about 4 years of optimizing applications and server environments for vm conversion, so I have a basis of experience to draw on.

i7

 

[Zstream]

Originally posted by: KentState

Originally posted by: Zstream

Originally posted by: aigomorla

Originally posted by: Zstream

Originally posted by: aigomorla
isnt VM ware based on parallism where number of threads actually play role?

if thats the case your telling the guy to go with a 4 threaded processor over a 8 threaded one.

You guys remember the i7 with HT that actually works can simulate 8 simulatanous threads.

Also 6 DDR3 slots vs 4.

*facepalm* Do any of you do actual server work?

As in VMWare? no but i was interested in setting one up on my gainestown.

Which is why im asking.

I Always thought more cores or thread application was better.

And thanks for the face palm.

More cores is better but I7 HT is still the same as P4 HT with slight modifications. Any many cases HT performs worse then nonHT. Such as in exchange, VM, sql 2005 etc..

VMware and HT perform better with little CPU load. If the VMware is above 25% load HT needs to be turned off.

Mr. Facepalm, I like how you state as a matter of fact that SQL Server performs worse under HT, when that is not necessarily the case. This should be tested by the organization for their particular instance. I've seen a substantial gain in performance using it with a data warehouse, various applications, and with a B2C website. Even MS states that a 10-20% improvement can be gained from using it.

http://support.microsoft.com/kb/322385

No, please reread what I stated, SQL 2005. Very few people use 2000..... or earlier.

 

[Zstream]

Originally posted by: i7guy1
Different strokes for different applications. Since you don't know what I manage, and I'm not saying, you can either take my comments at face value, or not. Either way it doesn't matter.

I have about 4 years of optimizing applications and server environments for vm conversion, so I have a basis of experience to draw on.

i7

I am sure you do with the name of your avatar it makes me believe you...

 

[classy]

You should look to purchase a used Dell 2850 or 2900/2950. And use VMware ESXi which installs on bare metal. It will be far better and for just study you shouldn't need more than 4gb ram anyway. I am also upgrading my MCSE 2K3 to 2K8 as well.

 

[Duvie]

Originally posted by: Zstream

Originally posted by: aigomorla
isnt VM ware based on parallism where number of threads actually play role?

if thats the case your telling the guy to go with a 4 threaded processor over a 8 threaded one.

You guys remember the i7 with HT that actually works can simulate 8 simulatanous threads.

Also 6 DDR3 slots vs 4.

*facepalm* Do any of you do actual server work?

Quit being a know-it-all punk.....

 

[bradley]

Obliterate is a pretty powerful word, perhaps even contentious if no attempt is made at proof.

 

[Zstream]

Originally posted by: classy
You should look to purchase a used Dell 2850 or 2900/2950. And use VMware ESXi which installs on bare metal. It will be far better and for just study you shouldn't need more than 4gb ram anyway. I am also upgrading my MCSE 2K3 to 2K8 as well.

Another vote for the 2900 series with two Intel Xeons X5470. I prefer the Opteron 8356 but it all comes down to price, the opteron is about 100% more powerful in Vmware then the Xeon but twice as expensive.

 

[classy]

Here is a list of ebay dell servers

Many have some excellent configurations already.

 

[ajaidevsingh]

I think the Sun Fire V20z is lovely.