p4 2.8 reviews are hitting the web!!

[Duvie]

Reviews are at

www.tomshardware.com
www.xbitlabs.com
www.hardocp.com

Here are my thought...



Lets point out a couple of things about Toms hardware overclocking section.

1) They hit 3325 with the 400mhz fsb becuase the rdram limits the oc and the rdram mobo couldn't go past 148fsb, so the 3108 was not the highest the 2.8ghz (533fsb) could hit...DDR mobos are likely to go higher and with higher cpu speed and 440+ mhz ddr can make up rdram advantage with a 100-150mhz more cpu speed....At that case the below numbers would show a 3.1ghz mobo with 440ddr should still beat an athlon 3400+ in majority of test done.

2) The new 2.8's take 1.525v...Not too bad when you think of the different vcores for the amd tbreds between the proposed lower tbreds 1600+ and the 2600+....

3) Done with conventional cooling...We all know the amd 3400+ was not conventional cooling by any means...


Now for some results....

Head comparisons of benchmark tests minus sysmark2002 and that lame rdram versus 333mhz ddr review....both cases I feel unfairly show the p4 or is not a good comparison...

Toms

p4 2.8 vs 2600+.............INtel 80% AMD 20%

p4 2.8 vs oc'd 3000+......Intel 64% AMD 36%

p4 2.8 vs oc'd 3400+......Intel 52% AMD 48%
Yodayoda where are you!!!!

p4 oc'd 3.1 vs 3400+......INtel 76% AMD 24%


So it basically appears the athlon shouldn't regain title again until hammer, but who knows if the 2800+ comes out with 333mhz fsb it could help but I think it still doesn't beat the 2.8 p4 not with these numbers.

Edit:

Xbits

p4 2.8 vs 2600+.......Intel 66% AMD 33%

minus sysmark2002 BS again...and note they used 333mhz ddr or basically the pc800 not pc1066 like toms...so ppl with 400mhz ddr and i850e platforms would likely fit more in line with toms 80/20.


Hardocp

only ran 9 test!!!

p4 2.8 vs 2600+.......Intel 89% AMD 11%

again 333mhz ddr on an p4b533-v mobo...

Oc'd the SOB to 3.3ghz at 1.7v...WOW....

 

[WarCon]

How did they get it to go over 3ghz without a new motherboard? With all the threads about requiring a new motherboard for over 3ghz, those numbers just can't be right.................




[The above was mild to medium level sarcasm, please take it as such]

 

[Duvie]

I did...I also noted in recent thread that many have had 2.533ghz in the 3+ ghz range for months now and no reported issues I have heard. Maybe since they are oc'd and vcore adjustments are made this is not a factor...I don't know!!!

 

[BD231]

Originally posted by: WarCon
How did they get it to go over 3ghz without a new motherboard? With all the threads about requiring a new motherboard for over 3ghz, those numbers just can't be right.................




[The above was mild to medium level sarcasm, please take it as such]

How you ask?, BS is how!!

 

[Duvie]

How you ask?, BS is how!!

I hope you are joking as well...Otherwise you are calling a lot of ppl liars at reputable sites and not to mention quite a few ppl around that have hit that with their 2.53ghz machines...

 

[ravedave]

aceshardware.com has an awsome article on the 2.8. It appears there were some un announced core changes to the p4, that give from 1-5% improvements over the same clock speed.

 

[FishTankX]

You're completly forgetting the barton and it's 333FSB and 512K L2 cache. What's the gap between the 2.0A and the 2.0? Quite serious. What's the gap between the 2.2 and the 2.26? Decent. Add these two together and suddenly the Athlon is significantly more competitive, espically with the synchronous CAS 2.0 DDR333 memory bus. The Barton should compete nicely with the P4 2.4 and perhaps even the 3.0. Did the Barton slip your mind, Duvie? My 1.7 Williamette is feeling old..

 

[Duvie]

Was that the added instruction sets???

 

[Duvie]

Nope....I agree the 512 could add some benefit and catch the 2.8ghz but the fact the 3400+ oc'd to 333fsb doesn't beat the 2.8 in majority of test does not bode well...Even if the 512 cache adds 10 percent like it did in most cases with the willamette versus the northwood...It wont make up enough distance to keep pace with the 2.933 and 3.066ghz chips....

 

[Wingznut]

FishTank... Well, you certainly cannot compare the P4's reaction to those enhancements with the Athlons. The same results will not be seen with the K7 architecture.

For example, The Tech Report recently had an article comparing the 133mhz vs 166mhz FSB on an Athlon... They concluded that the vast majority of results showed a negligible delta between the two.

Besides... Who knows if Barton will show up before Hammer, if it shows up at all?

 

[Leon]

Tom needs to update his "price comparison"

 

[FishTankX]

You realize Duvie that the bus overclock on the P4 was a bit more extreme than the bus overclock on the thoroughbred, combine that with the fact that you'll have a reduced multiplier with the AthlonXP (which will help it scale infinitley better) and more cache (I must admit, the Thoroughbred's cache, and the Athlon's cache in general is pretty sad though) will keep it competitive if they price it well enough. Note:I said competitive! Not necescarily beating it, but it would keep it within 3 or 4% of it at max. Imagine a 2.4GHZ Athlon with 512K cache and 333FSB. And also, Duvie, if I may remind you, we have *never* seen a cache bump on the AthlonXP architecture. So you have no idea how much it might increase performance. The cache bump on the Pentium3 resulted in an obscene increase in performance, although it's hard to say how much of that performance increase was atributed to the new prefetch logic.

I think if they would just merely bump up the bus width on the L2 cache the Athlon could really roar!

Cache bandwidth
P4 2.53GHZ(256 bit cache bus) =~80GB/s with SSE2 enabled
AthlonXP 2.53GHZ(64 bit cache bus)=~10GB/s I believe.

Big difference, huh?

 

[Duvie]

I agree with wings...The talk is the 333mhz fsb may come with the tbred chips and not the barton now, and I heard nothing about 512kb cache then....I also think the timing is getting to close to the hammer and in all likelihood it wont be here early enough to do what you want it to do or compare to what you want it to....

It remains to be see non the cache you are right there...Wings is also right and many amd users have commented no big difference using the 333 mhz fsb. So with no real delta there while the p4 had some nice improvement in some apps the cache better be worth more to it.

Edit: I would have to say if the 512 l2 cache would make a huge factor in performance I truly believe AMD would have implemented it already in the past and erase the lackluster performnace in the last 1/2 year. I am sure they know its benefit and if it was s great as to keep the title while they produced the hammer and offset the added cost they would have done it....Now the hammer is totally different, right?? It doesn't have need for that much l2 cache, right??

 

[Wingznut]

Originally posted by: Duvie
The talk is the 333mhz fsb may come with the tbred chips and not the barton now...

The next Athlon announcement (2200mhz?) will probably include a 166mhz FSB.

 

[andreasl]

The Instruction TLB was increased from 64 to 128 entries. If you want to read a good explanation about what a TLB is, check out this post over at RWT:

http://www.realworldtech.com/forums/index.cfm?action=detail&PostNum=920&Thread=4&entryID=9800&roomID=11