3DMark Vantage Loves The Cores: QCs FTW !!

[Lightman]

But a Q9300 has LESS cache than a Q6600. So it aint that (3mb vs 4Mb)

It does have larger associativity though, and less latency iirc.

also the FSB is higher.. a possibility

Good point

My theory is that 3DMV uses SSE4 for CPU test/tests or most recent PhysiX library is using it. This way we can explain very big difference between Kentsfield and Yorkfield at similar clocks.

[Syn.]

How is a 9850 at 3.0ghz getting beat by a 2.5ghz Q9300 showing off Phenom's competitiveness?

Lol, I'll probably get flamed for this, but I'm just pointing out the obvious.

I was thinking the same thing. I guess the only way is that 9850 costs less then Q9300, about £20 less.... nope that aint either.

Why wont people just accept Phenom for what it is?

[perkam]

I was thinking the same thing. I guess the only way is that 9850 costs less then Q9300, about £20 less.... nope that aint either.

Why wont people just accept Phenom for what it is?

Please stick to the topic and the first post.

The Phenom is a Work in Progress from AMD (the TLB issue proved that) and, as per the content of this thread, gains a slight advantage over previous benchmarks, hence the thread.

Think of it this way, good news is rare for AMD, hence this is rare news :p Every AMD thread in XS News doesn't need members to brand the entire Phenom line useless over and over again. Thankfully we have the vacancies for doing that filled 200 times over.

Perkam

[villa1n]

Im still wondering about when you increase the clocks on a q9300 20% how you get more than a 20% increase... lol, it should be the other way around...

2.5ghz to (20%) 3.0 ghz.

10041 pts to (20.6%) 12106 thats straight line scaling!

[gallag]

Please stick to the topic and the first post.

Just to clarify, do you want to stick to the original first post or the edited one? Just getting a bit confusing.

[saaya]

yeah the 9850 is nice... cheaper than a q6600 and only slightly slower
if your ocing... forget about it... but otherwise, a nice quadcore...

then again who needs a quadcore?
amd should have focussed on a really nice dualcore cpu instead of a "monolithic" quadcore cpu that is oh so efficient...

[saaya]

just look at the results, vantage is very cache friendly....

a phenom +0.1 speed increase provides roughly + 250/280 score depending on x3 or x4
same for the q9300-q9450 comparison + 278 points

but going from 7300-8300 @ same speed you increase more then 600 points
so increasing from q6600-q9300 gives you 0.1 speed advantage + 2x cache increase so that explains the big jump.

nice analyzing!

[Periander6]

yeah the 9850 is nice... cheaper than a q6600 and only slightly slower
if your ocing... forget about it... but otherwise, a nice quadcore...

Cheaper? You see the 9850 for less than $219?

http://www.newegg.com/Product/Produc...82E16819115017

http://www.newegg.com/Product/Produc...82E16819103249

Try slower and more expensive. Also poorer overclocking and higher power consumption. How is this better than Prescott?

Please stick to the topic and the first post.

The Phenom is a Work in Progress from AMD (the TLB issue proved that) and, as per the content of this thread, gains a slight advantage over previous benchmarks, hence the thread.

Think of it this way, good news is rare for AMD, hence this is rare news :p Every AMD thread in XS News doesn't need members to brand the entire Phenom line useless over and over again. Thankfully we have that vacancies for doing that filled 200 times over.

Perkam

New life? More like still stone dead. There is no good news here for AMD. Yes, Phenom is a work in progress, and Intel's processors are not? You post on an overclocking website a chart that shows the top bin max overclock Phenom that just came out getting crushed by 40% vs a bottom bin Ye Olde Kentsfield overclock and want people to "stick to the topic" of all the "new life" in Phenom?? What is this, AMDZone?

[Epsilon84]

Im still wondering about when you increase the clocks on a q9300 20% how you get more than a 20% increase... lol, it should be the other way around...

2.5ghz to (20%) 3.0 ghz.

10041 pts to (20.6%) 12106 thats straight line scaling!

One possible explanation is the FSB overclocking. For some reason the default multi (7.5x) was not used and they used 7x instead, resulting in a higher FSB to clock ratio, which could potentially explain the extra 0.6%.

[perkam]

Cheaper? You see the 9850 for less than $219?

http://www.newegg.com/Product/Produc...82E16819115017

http://www.newegg.com/Product/Produc...82E16819103249

Try slower and more expensive. Also poorer overclocking and higher power consumption. How is this better than Prescott?




New life? More like still stone dead. There is no good news here for AMD. Yes, Phenom is a work in progress, and Intel's processors are not? You post on an overclocking website a chart that shows the top bin max overclock Phenom that just came out getting crushed by 40% vs a bottom bin Ye Olde Kentsfield overclock and want people to "stick to the topic" of all the "new life" in Phenom?? What is this, AMDZone?

For starters, most phenom 9850s clock to 3ghz, not just the top bin ones, thanks to AMD per core clocking utility and unlocked mutli.

Secondly, I noticed the word vantage did not appear once in your post So my advice on sticking to the topic is valid. I am not going to repeat my points on why I believe the Phenom has gained an advantage. Please look it up in previous pages or in the first post and respond accordingly

I am wondering if anyone has a similar chart for the remaining CPU tests in 3DM Vantage. If anyone has seen them pls post them here I am looking forward to more Phenom rigs on the ORB to see how people make use of them

I have changed the format and title of the thread, however, as I`m starting to get a bad case of deja vu from the tone of your post

Perkam

[Syn.]

Please stick to the topic and the first post.

The Phenom is a Work in Progress from AMD (the TLB issue proved that) and, as per the content of this thread, gains a slight advantage over previous benchmarks, hence the thread.

Think of it this way, good news is rare for AMD, hence this is rare news :p Every AMD thread in XS News doesn't need members to brand the entire Phenom line useless over and over again. Thankfully we have the vacancies for doing that filled 200 times over.

Perkam

I was sticking to the topic you first posted.

All processors are work in progress. If you read any of my posts or if you just read my signature you would know that i am not against Phenom. Its just that this thread seemed as if someone invented hot water and then decided to tell the whole world about it, if you get what i mean. Currently with B3's a Phenom is a very decent CPU the only problem here being is that currently it is not decent enough because Intel released something that performs better before hand. As you know one thing that has been proven many times over, especially in XS News, is that Phenom is slower clock-per-clock then Core Quad in most cases. So if you really want people to let this pass you should not come out with threads that say "aha but now take a look this..." and reopen the discussion that people are mostly annoyed of reading these days.

So my point is this. Phenom is a good cpu to have, just because its not best is not the end of the world it will still run all my applications and games without any problems.

[Epsilon84]

For starters, most phenom 9850s clock to 3ghz, not just the top bin ones, thanks to AMD per core clocking utility and unlocked mutli.

Phenom 9850 *is* a top bin CPU, do you not understand the concept of CPU binning?

Anyway, I think you should take it easy with your uber edits, it takes the discussion COMPLETELY out of context.

[grimREEFER]

wow, the 9850 at stock is faster than the e8500@ 3.8ghz by a huge amount.
well, that makes sense, the e8500 just costs $20 more...
intel's high end dual core cpu's are the biggest rip off, intel's low end quad cores are the best value.

[stocius]

Pointless fanboyism. How's this. I don't give a either way. Both companies suck. I want 10 GHz on stock cooling. Wake me up when that gets here.

[Epsilon84]

wow, the 9850 at stock is faster than the e8500@ 3.8ghz by a huge amount.
well, that makes sense, the e8500 just costs $20 more...
intel's high end dual core cpu's are the biggest rip off, intel's low end quad cores are the best value.

If 3DMark is your life...

There are plenty of instances where a higher clocked dual core can outperform a lower clocked quad:

http://www.extremetech.com/article2/...2282399,00.asp

[WesM63]

How does 8 cores do?

[Exedy]

3DMarketing could be renamed as CPUMarketing

[villa1n]

One possible explanation is the FSB overclocking. For some reason the default multi (7.5x) was not used and they used 7x instead, resulting in a higher FSB to clock ratio, which could potentially explain the extra 0.6%.

true, i guess the bandwidth would be opened up from 333, to 400+ they needed.
was it ddr2 for both, or ddr2 and ddr3? .. not really clear or didnt see a link to the test beds, which could make a big diff.

[Mungri]

Pointless fanboyism. How's this. I don't give a either way. Both companies suck. I want 10 GHz on stock cooling. Wake me up when that gets here.

I'm so agreeing with this post and quoting it for that reason :p

And no, I dont think I actually need a quad core yet anyway. Maybe by the time 12 core CPU's are released, games will start being written to take advantage of 4 cores

[JumpingJack]

yeah the 9850 is nice... cheaper than a q6600 and only slightly slower
if your ocing... forget about it... but otherwise, a nice quadcore...

then again who needs a quadcore?
amd should have focussed on a really nice dualcore cpu instead of a "monolithic" quadcore cpu that is oh so efficient...

http://computers.pricegrabber.com/pr...9850/st=query/

http://computers.pricegrabber.com/pr...6600/st=query/

Actually, they are about the same, the Q6600 looks to be a bit less on average.... just taking the lowest three prices from pricegrabber

Q6600: 219, 220(219.99), 226 (225.99)
9850 : 235, 236(235.99), 236(235.99)
(As of 4/30/08)

[JumpingJack]

For starters, most phenom 9850s clock to 3ghz, not just the top bin ones, thanks to AMD per core clocking utility and unlocked mutli.

Perkam

Perkam... the 9850 is AMD's top bin ... they don't bin out anything faster. EDIT: At the moment, 2.6G will arrive later per their roadmap.

[gallag]

penryn breathes new life into specweb2005

Top scores (In parenthesis - subtests scores: Banking(SSL) / Ecommerce(partially SSL) / Support (non-SSL))


HP ProLiant DL580 G5, 4S/16C/16T, Intel Xeon X7350, 2930 MHz, 64GB, 58 HDs, RHEL5, Accoria Rock Web Server v1.4.6 - 40046 (71104/55552/36032)
Sun SPARC Enterprise T5220, 1S/8C/64T, Sun UltraSPARC T2, 1400 MHz, 64GB, 50 HDs, Solaris 10, Sun Java[TM] System Web Server 7.0 Update 2 - 37001 (63000/49500/36000)

HP ProLiant DL385 G5, 2S/8C/8T, AMD Opteron 2356, 2300 MHz, 32GB, 51 HDs, RHEL5, Accoria Rock Web Server, v1.4.6 - 30007 (50856/46020/25584)

HP ProLiant DL380 G5, 2S/8C/8T, Intel Xeon X5460, 3166 MHz, 32GB, 58 HDs, RHEL5, Accoria Rock Web Server, v1.4.6 - 29591 (51840/46512/23816)


Look at those Intel chips rock and finally position them selfs as a strong alternative to AMD in high bandwidth applications.

I realize this post will provoke strong debate because Intel actually fairs quite poorly but that does not matter as I can always edit my first post. After all I am just trying to prove that I can not have a positive Intel thread without AMD fans hijacking it.

[villa1n]

penryn breathes new life into specweb2005

Top scores (In parenthesis - subtests scores: Banking(SSL) / Ecommerce(partially SSL) / Support (non-SSL))


HP ProLiant DL580 G5, 4S/16C/16T, Intel Xeon X7350, 2930 MHz, 64GB, 58 HDs, RHEL5, Accoria Rock Web Server v1.4.6 - 40046 (71104/55552/36032)
Sun SPARC Enterprise T5220, 1S/8C/64T, Sun UltraSPARC T2, 1400 MHz, 64GB, 50 HDs, Solaris 10, Sun Java[TM] System Web Server 7.0 Update 2 - 37001 (63000/49500/36000)

HP ProLiant DL385 G5, 2S/8C/8T, AMD Opteron 2356, 2300 MHz, 32GB, 51 HDs, RHEL5, Accoria Rock Web Server, v1.4.6 - 30007 (50856/46020/25584)

HP ProLiant DL380 G5, 2S/8C/8T, Intel Xeon X5460, 3166 MHz, 32GB, 58 HDs, RHEL5, Accoria Rock Web Server, v1.4.6 - 29591 (51840/46512/23816)


Look at those Intel chips rock and finally position them selfs as a strong alternative to AMD in high bandwidth applications.

I realize this post will provoke strong debate because Intel actually fairs quite poorly but that does not matter as I can always edit my first post. After all I am just trying to prove that I can not have a positive Intel thread without AMD fans hijacking it.

this isnt really related to this thread at all, i think you meant to post in the spec web thread....

also.. its total bait. the intel server there is clocked 800mhz more.. and still underperforms... this is not their forte yet.

[Epsilon84]

this isnt really related to this thread at all, i think you meant to post in the spec web thread....

also.. its total bait. the intel server there is clocked 800mhz more.. and still underperforms... this is not their forte yet.

I guess sarcasm doesn't convey well over the Internet huh?

[Budwise]

they dont bother to post a high clocked E8500 or 8400...