Intel Xeon 5570: Smashing SAP records.

[GoriLLakoS]

Where we can find this specific benchmark?

[gosh]

LOL keep trying gosh.

Trying?

You should be suspicious if a company present numbers on their products or if the data presented are strange and with sparse information. Doesn't matter if it is AMD or Intel.
If a company presents numbers and also is selling products that it presents numbers on, most will be suspicious.

[Clairvoyant129]

Trying?

You should be suspicions if a company present numbers on their products or if the data presented are strange and with sparse information. Doesn't matter if it is AMD or Intel.
If a company presents numbers and also is selling products that it presents numbers on, most will be suspicious.

Yea right, if it was the other way around, you would be the first to post and comment, "new AMD Opterons destroy Intel CPU!!!!"

[gosh]

Yea right, if it was the other way around, you would be the first to post and comment, "new AMD Opterons destroy Intel CPU!!!!"

Wrong!

[Movieman]

Yea right, if it was the other way around, you would be the first to post and comment, "new AMD Opterons destroy Intel CPU!!!!"

Didn't I just ask that we chill out a little?
I'm sitting here in lovely New Hampshire where my house has been without electrical power since last Thursday running off a 5500w generator since Sunday and delivering a whole 30 amps to my house.
My next door neighbor lost his cold water pipes and is looking at $5-10G's to repair.
Let's keep a perspective here ok.
This is news in that it's first info and should be taken as maybe right and maybe off by a country mile but definaely not something to argue over.

[savantu]

Where we can find this specific benchmark?

http://www.sap.com/solutions/benchmark/sd.epx

As you know , SAP is one of largest SW companies in the world.

Didn't I just ask that we chill out a little?
I'm sitting here in lovely New Hampshire where my house has been without electrical power since last Thursday running off a 5500w generator since Sunday and delivering a whole 30 amps to my house.
My next door neighbor lost his cold water pipes and is looking at $5-10G's to repair.
Let's keep a perspective here ok.
This is news in that it's first info and should be taken as maybe right and maybe off by a country mile but definaely not something to argue over.

Well , let's not make it personal , shall we ?

You live in New Hampshire and have your own set of problems , I live 7000 miles from you in Romania , someone else lives in Santiago , Chile , etc.Each with his own set of problems , which we don't move online.And no , the US isn't centrum mundi.

Gosh's track record entitles people to doubt his neutrality ( as if he had one ) and not to give him a carte blanche.He's in the same region as kassler or duby229 for those in the know.

[Ghostbuster]

Where we can find this specific benchmark?

Check this http://www.sap.com/solutions/benchma...ier.epx?num=20

You should be suspicious if a company present numbers on their products or if the data presented are strange and with sparse information. Doesn't matter if it is AMD or Intel.
If a company presents numbers and also is selling products that it presents numbers on, most will be suspicious.

You're trying too hard, gosh... but you will always get pie'd all over your face no matter what. There's nothing wrong with a company trying to promote their products, and given the economic climate they would certainly like attact and to "lock" in potential clients to buy their systems. Imagine 2 socket quad core system faster than a 4 socket quad core system, which will be considerably cheaper and lower maintenance (less power consumption).

[Ghostbuster]

Gosh's track record entitles people to doubt his neutrality ( as if he had one ) and not to give him a carte blanche.He's in the same region as kassler or duby229 for those in the know.

Gosh and kassler is the same person...

[Movieman]

Well , let's not make it personal , shall we ?

You live in New Hampshire and have your own set of problems , I live 7000 miles from you in Romania , someone else lives in Santiago , Chile , etc.Each with his own set of problems , which we don't move online.And no , the US isn't centrum mundi.

Gosh's track record entitles people to doubt his neutrality ( as if he had one ) and not to give him a carte blanche.He's in the same region as kassler or duby229 for those in the know.

You misunderstood me. What I was trying to say that in the big picture of life this deal over the cpu's is a small thing and not worth arguing about.
Yes,I know that people here love one company or the other so at times their comments may be slanted but take it for what it is..Not that important.

[Shintai]

Intel is on quite a track record.

First they conroe the desktop/mobile market. Then they "conroe" the SSD market. And now they do it again in the server segment.

Its no surprise for those tried both i7 and Xeon 5500 series to know that this is another revolution in the workstation/server segment.

Quad socket and then some in a cheap dual socket 2U system. Its basicly a 2-4000$ now instead of a 10000$+ server.

[gosh]

Gosh's track record entitles people to doubt his neutrality ( as if he had one ) and not to give him a carte blanche.He's in the same region as kassler or duby229 for those in the know.

Did you read this?

• Sensitive to Sync (“cache coherency”) latency

One of the biggest bottlenecks for Intel has been the sync latency. It is possible that once the "sync" bottleneck was removed

In another thread (GTA IV) I informed about this and that wasn't popular because it is very hard to criticize Core 2 for the not having a L3 cache and data needs to travel through FSB which leads to bad synchronization between threads.
I think that you will see more applications that will gain much more now because you finally can take advantage of more cores without having to think about the design of the Core 2 processors.
About this test I don't know how it is done, what I do know is that if you run databases there are a lot of things in the computer hardware that has effect on performance.
When it comes to servers it is no problem to talk about synchronization but on desktop's some go crazy.

[Shintai]

Did you read this?

• Sensitive to Sync (“cache coherency”) latency

One of the biggest bottlenecks for Intel has been the sync latency. It is possible that once the "sync" bottleneck was removed

In another thread (GTA IV) I informed about this and that wasn't popular because it is very hard to criticize Core 2 for the not having a L3 cache and data needs to travel through FSB which leads to bad synchronization between threads.
I think that you will see more applications that will gain much more now because you finally can take advantage of more cores without having to think about the design of the Core 2 processors.
About this test I don't know how it is done, what I do know is that if you run databases there are a lot of things in the computer hardware that has effect on performance.
When it comes to servers it is no problem to talk about synchronization but on desktop's some go crazy.

So according to you. Hypertransport stinks at this?

i7 already showed it can match dualsocket systems. So your statements are quite uneducated to put it mildly.

[Movieman]

Intel is on quite a track record.

First they conroe the desktop/mobile market. Then they "conroe" the SSD market. And now they do it again in the server segment.

Its no surprise for those tried both i7 and Xeon 5500 series to know that this is another revolution in the workstation/server segment.

Quad socket and then some in a cheap dual socket 2U system. Its basicly a 2-4000$ now instead of a 10000$+ server.

And if those numbers ARE true in the pic posted on that site remember that they are using X5570's at 2930.
A buddy has some X5580's.... Just a little faster at 3070mhz

[gosh]

So according to you. Hypertransport stinks at this?

i7 already showed it can match dualsocket systems. So your statements are quite uneducated to put it mildly.

I think AMD still has hypertransport I on servers. Was turbo mode enabled on i7
What was the speed of the memory and how much memory was used? Does harddrive performance have anything to do with this test? Which compiler was used to compile the test?

[Slay0r]

Good on Intel and on the server market as a whole if true. Eager to see these in action.

Gosh, can you just log out and come back later on? Let's say when you stop trying to defend perfectly fine companies that really could do without people with apparent subzero IQ to wave weapons made of whine in their defence when they really don't need it.

[Donnie27]

Did you read this?

• Sensitive to Sync (“cache coherency”) latency

One of the biggest bottlenecks for Intel has been the sync latency. It is possible that once the "sync" bottleneck was removed

In another thread (GTA IV) I informed about this and that wasn't popular because it is very hard to criticize Core 2 for the not having a L3 cache and data needs to travel through FSB which leads to bad synchronization between threads.
I think that you will see more applications that will gain much more now because you finally can take advantage of more cores without having to think about the design of the Core 2 processors.
About this test I don't know how it is done, what I do know is that if you run databases there are a lot of things in the computer hardware that has effect on performance.
When it comes to servers it is no problem to talk about synchronization but on desktop's some go crazy.

You do know that both companies you accusing of improprieties are AMD partners, right? HP sells a crap loads of AMD servers and always has. HP turned down Intel Rebates to sell AMD servers, workstations and desktops. Then still made more money than Intel only at that time Dell. So now you're saying they're lying for Xeons to make Intel look better? HP is an Intel partner but not its yes man or sucker. Through the years they've more than proven that.

Trying?

You should be suspicious if a company present numbers on their products or if the data presented are strange and with sparse information. Doesn't matter if it is AMD or Intel.
If a company presents numbers and also is selling products that it presents numbers on, most will be suspicious.

Don't remember you being suspecious of the AMD saying all of their first Phenoms would reach 3GHz or the 6GHz LN stuff

Oh my gosh! Pardon the pun. Without bashing you but just disagreeing with your comments, there was always clues this was going to happening.

Intel's FSB's was the biggest bottleneck and was Xeon's biggest draw back. The first Xeon based of the New Core architecture, Showed this going from two FSB's to 4 Cores with Two (separate dual) FSB's. It was handicapped to say the least and even the old ones would have done a lot better on an Alpha EV6 or 7 type BUS.

Then add in all of the Core improvements and yes, nothing strange or questionable at all LOL! AMD had been winning Apache Web Srv apps tests since the first K7 showed up, yet, Opteron got spanked. Many members here KNOW point to point killz the FSB. Put the older Xeons on a Similar Alpha EV6 or 7 and it would do a lot better. But as usual, instead of listening, many folks saying this were called Intel Fans and dismissed. Now add all the things Intel did to make it do Parallel Processing better and Wa-La.

Note, as well as it has been said all over the web that most analyst think Nehalem was made for server first then Desktop and notebooks as an afterthought. How did folks miss this being repeated over and over again.

[justapost]

Do those Xeos have the turbo feature? How many multis are possible if? Sorry I'll be impressied if they show those results in the spec_power benchmark. I'm sure those results are tuned for marketing reasons as they always are.
Once those chips are available amd can close the gap with six core instanbul chips.

[Donnie27]

Do those Xeos have the turbo feature? How many multis are possible if? Sorry I'll be impressied if they show those results in the spec_power benchmark. I'm sure those results are tuned for marketing reasons as they always are.

But isn't that like asking was L3 disabled or something? I hope all features were on, it proves it works.

[Shintai]

Do those Xeos have the turbo feature? How many multis are possible if? Sorry I'll be impressied if they show those results in the spec_power benchmark. I'm sure those results are tuned for marketing reasons as they always are.

Yes they do. Its the standard turbo mode. as in 133 and 266mhz above. In this case it would be 133mhz above. If TDP allowed it.

And DDR3-1333 memory and 6.4GT QPI..

These aint tuned for marketing. Simply look on i7 reviews. Single i7 often beat or get close to a skulltrail system with Q9775. And it just gets worse with scaling since QPi is the new kid on the block.

[gosh]

You do know that both companies you accusing of improprieties is an AMD partner, right? HP sells a crap loads of AMD servers and always has. HP turned down Intel Rebates to sell AMD servers, workstations and desktops. Then still made more money than Intel only at that time Dell. So now you're saying they're lying for Xeons to make Intel look better? HP is an Intel partner but not its yes man or sucker. Through the years they've more than proven that.

My critizism (If I had some because the first was a simple joke) is for how Anandtech presented the facts. i7 is a very fast processor, it is almost like that they have done it to be good for databases. It is a bandwidth monster. But they applaud Intel for having L3 cache and other things used to improve multithreading and in the same article they say that AMD is bad knowing that they was first with L3 and hypertransport (QPI for intel). I don't know how much more you can prove that you isn't neutral.
The facts in the article was way to sparse to do any conclusions what I think. It is no brainer though that i7 will get very good performance running databases. I also think they have some new instructions for parsing data or if some instructions have improved for this (don't remember)

[justapost]

Yes they do. Its the standard turbo mode. as in 133 and 266mhz above. In this case it would be 133mhz above. If TDP allowed it.

So it's like on the locked desktp chips and not like on the i965 where you can go up more than one multi on all cores?
OT (because it's about performance only here) TDP is good for comparison agains xeos but for platform comparisons you can not rely on that value because all the uncore parts are not taken into account. Also I really expect high temperatures in those i7 servers.

And DDR3-1333 memory and 6.4GT QPI..

Hmm that may have an big impact since it's DDR2-800 and 4GT HT2 on amd side atm. Current phenoms have a bandwidth (sandra) around 10.5-11GB/s with DDR2-1066. Some preview's mentioned ~14GB/s for PII's with DDR3. Might still fall behind nehlames where whom get this bandwidth with ddr3-1055 cl5. HT3.1 will come 2010 and sixcores will shorten the gap in 2009.

These aint tuned for marketing. Simply look on i7 reviews. Single i7 often beat or get close to a skulltrail system with Q9775. And it just gets worse with scaling since QPi is the new kid on the block.

geekfool showed an good example how well SAP is tunable on the anandtech's report comment section.

I have a friend who used to run SAP SD 2-tier benchmarks for his employer. This benchmark is well known to yield very different results depending on how the system is tuned. Here are 2 examples of 2 pretty much identical systems (dual-Opteron 2356, 32 GB RAM, exact same software version), with one achieving a 86% higher SAPS value due to different undisclosed and unreported tunings made to the 2nd system:

5730 SAPS: http://download.sap.com/download.epd...A4007AC4A4875E
10520 SAPS: http://download.sap.com/download.epd...3CE500D893C1BB

Because of that, absolutely nothing can be extrapolated from the "surprisingly" high Xeon W5570 result... Now all the engineers working on the teams submitting the results know what are causing them to vary so much, but expect the marketing departements to claim this is all due to how their servers are so much better than their competitors'.

With an 43% deviation this benchmark doe not look very usefull. I like i7 from architecture perspective. High wattage for the cpu is ok as long as perf/watt benefits and temps do not exceed. I expect temps to be the bigges issue for i7 in server space.

[Jacky]

Once those chips are available amd can close the gap with six core instanbul chips.

Yeah, which will appear out of thin air to save them? Or will it rather launch in Q4'09 competing against Beckton (8 cores, 24mb L3, quad channel memory) and Westmere (32nm Nehalem)?

geekfool showed an good example how well SAP is tunable on the anandtech's report comment section.

With an 43% deviation this benchmark doe not look very usefull. I like i7 from architecture perspective. High wattage for the cpu is ok as long as perf/watt benefits and temps do not exceed. I expect temps to be the bigges issue for i7 in server space.

He is clueless as per Michael S from realwordtech (which BTW I have already linked in my first post) and I would believe a regular RWT poster any day over some random guy bashing anandtech in their comments, which unfortunately is happening all the time.

[Tiridum]

Might still fall behind nehlames where whom get this bandwidth with ddr3-1055 cl5. HT3.1 will come 2010 and sixcores will shorten the gap in 2009.

More cores will not solve this problem, this is about bandwidth, more cores will likely result in about similar performance then quadcores because the cores are bandwidth starved

[Shintai]

geekfool showed an good example how well SAP is tunable on the anandtech's report comment section.

Not really. He just failed completely in reading the numbers.
I think you should read the original info again.

Basically, there are two classes of SAP-SD 2-tier submissions - "fast" with response time around 1 second and "throughput-oriented" with response time around 1.6-2 seconds. It seems, average response time above 2 seconds is illegal.

[Ghostbuster]

I like i7 from architecture perspective. High wattage for the cpu is ok as long as perf/watt benefits and temps do not exceed. I expect temps to be the bigges issue for i7 in server space.

Nopes, have you ever seen those server rack machines. Talking about air cooling, servers have the best and the noisiest (fans fans everywhere, front to back). 2 processors isn't much effort to keep cool, but 4 processors maybe..