• We should now be fully online following an overnight outage. Apologies for any inconvenience, we do not expect there to be any further issues.

939 or 754

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Avalon

Avalon

Diamond Member
Jul 16, 2001
7,571
178
106
Originally posted by: Duvie
i hear ya avalon...

however 1-5% for the cache, and 3-5% for the dual channel controller...it all starts to add up.....

You are right it cant be beat for budget...it is the mobile barton of right now.....
Click to expand...

I'm hoping that having less cache (thus less heat) will let me overclock further than a winchester normally could, to help negate that 10% in performance losses. I actually got my backplate in from Zalman today. I'm going to install it tomorrow right around noon time. I should be up and ready to go with my DFI, benching away by early afternoon. I must complain about the northbridge cooler, though. It's a dinky passive aluminum piece. I tried to remove it several nights ago, but was unsuccessful. I'm going to try again so I can apply some AS5 on the NF3 chip. It's going to need it.
 
C

clarkey01

Diamond Member
Feb 4, 2004
3,419
1
0
Originally posted by: mhillary
939 is the best
Click to expand...

LOL erm. Welcome to the forums.


realredpanda:

If I were id go S745, I always change my MB everytime I upgrade ( I wait so long that theres no real point in upgrading the CPU up to the motherboards limit as its still slow) E.G my 2000AXP needs replacing, and my MB only allows for a 2100+. So I have to get another MB everytime.

If that S745 is a Newcastle then get it, its a peech ! (2.4 Ghz, same speed as the 4000+, only missing 512k cache and only has one memory controller)
 
S

Shenkoa

Golden Member
Jul 27, 2004
1,707
0
0
Socket 754 is not the future of next gen AMD A64's but it will be just fine for an upgrade. Its a much better value then socket 939 at the moment. Plus, the next time you will want to upgrade, their will probally be a newer and better socket.
 
C

clarkey01

Diamond Member
Feb 4, 2004
3,419
1
0
Originally posted by: Wicked2010
clarkey, too bad no one taught you computer architecture.

It's ... not a peech.
Click to expand...

peach*


What you talking about ?> a 3400+ Newcastle will beat a 3500+ winnie, I was only saying that the 3400+ S745 was a great pick due to its clock speed of 2.4 Ghz (not too far behind the 4000+ in benchmarks)

 
W

Wicked2010

Member
Feb 22, 2005
123
0
0
A 3400+ NewCastle will NOT beat a 3500+ Winnie. Wake up and smell the coffee.

And... your quote is nonsense.
 
H

hertz9753

Member
Oct 13, 2004
35
0
0
The 3400 Newcastle will not beat the 3500 Winchester but it does tie with it. It loses to it in some things and wins in others. Read the reviews.
 
C

clarkey01

Diamond Member
Feb 4, 2004
3,419
1
0
2.4 Ghz, 512K cache, single channel Vs 2.2 Ghz, 512 K cache, dual channel.

Dual channel offers 3% more performance.
 
D

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
one thing to notice is we throw out the 1-5% claim that the dual channel makes all the time but look at the 2 out of the 3 games as well as the TMPGenc...here a 2.2ghz cpu beat the 2.4ghz 3400+....so therefore DC memory most have accouted for 8-9% gain.....I thought you ppl loved gaming here???

Also notice ow the 90nm was better then the 130nm by an average of 1%...what makes that increase considering biothe the 3500+ variations are DC memory???



I agree the 3400+ is hands down the best bang for the buck....the 3700+ may be the worst...

AMD are morons with their PR rating...

How in the fvck can this happen

2.4ghz 130nm 512kb 754 = 3400+
2.4ghz 130nm 1mb 754 = 3700+ *** doesn't deliver***
2.4ghz 130nm 512kb 939 = 3800+
2.4ghz 130nm 1mb 939 = 4000+

Absolutely ridiculous....600 pr points or 25% for DC, 512kb more of cache...I dont think so!!!!

the 3400+ was an Intel killer destroying the 3.4ghz but the 4000+ doesn't do much better...not 600mhz better at least...p rating is whacke as usual...
 
C

clarkey01

Diamond Member
Feb 4, 2004
3,419
1
0
Originally posted by: Zebo
Originally posted by: hertz9753
The 3400 Newcastle will not beat the 3500 Winchester but it does tie with it. It loses to it in some things and wins in others. Read the reviews.
Click to expand...

In almost every test it will. Here The $195 skt 754 3400 beats the $260 skt 939 3500 in every single benchmark.

http://www.behardware.com/art/imprimer/531/
Click to expand...

There ya go. So much for your put down comments Wicked2010, I watch the CPU scene a little more then you do..
 
Zebo

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
Originally posted by: Duvie
one thing to notice is we throw out the 1-5% claim that the dual channel makes all the time but look at the 2 out of the 3 games as well as the TMPGenc...here a 2.2ghz cpu beat the 2.4ghz 3400+....so therefore DC memory most have accouted for 8-9% gain.....I thought you ppl loved gaming here???

Also notice ow the 90nm was better then the 130nm by an average of 1%...what makes that increase considering biothe the 3500+ variations are DC memory???



I agree the 3400+ is hands down the best bang for the buck....the 3700+ may be the worst...

AMD are morons with their PR rating...

How in the fvck can this happen

2.4ghz 130nm 512kb 754 = 3400+
2.4ghz 130nm 1mb 754 = 3700+ *** doesn't deliver***
2.4ghz 130nm 512kb 939 = 3800+
2.4ghz 130nm 1mb 939 = 4000+

Absolutely ridiculous....600 pr points or 25% for DC, 512kb more of cache...I dont think so!!!!

the 3400+ was an Intel killer destroying the 3.4ghz but the 4000+ doesn't do much better...not 600mhz better at least...p rating is whacke as usual...
Click to expand...

The only real problem is 3400 is "underrated" and should have been called the 3600+.. they are holding that spot the a 2.2 1MB lvl2 DC chip though.

Why? Because A64 is bascially 1.5x faster clock for clock than p4. Then you add 5% for Dual channel and 5% for 1 MB lvl2.

So 3400+ = 2400Mhz x 1.5 = 3600+ and that's about right.

3700+? *** doesn't deliver*** = not true = 2400Mhz x 1.5 x 1.05 = 3780 Mhz (anand even comments in 3700 review how it's just as fast as 3800)

3800+? 2400Mhz x 1.5 x 1.05 = 3780 Mhz = close enough

4000+? 2400Mhz x 1.5 x 1.05 x 1.05 = 3969 Mhz = close enough
 
D

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
I think the 3700+ numbers are exasperated by the fact the 3400+ in underrated....Look at those test and se how the 3400-3700 was within each other marginally the whole way...not worth 300 pr points...
 
Zebo

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
Originally posted by: Duvie
I think the 3700+ numbers are exasperated by the fact the 3400+ in underrated....Look at those test and se how the 3400-3700 was within each other marginally the whole way...not worth 300 pr points...
Click to expand...

Most are overrated as I showed... the 3400 and 3200 NC are exceptions and are underrated which i think skews things..

Some other examples of overrated include the 2800 which should have been called the 2750 but I guess AMD shys away from odd numbering as well as odd numbers so 2700 and 2750 are not used.

The 3500 should be a 3450.
The 3000 winny should have been called 2835...

Just look at various benchmarks you'll see my math borne out true..

Gloabally speaking A 2800 is a bit slower than a 2.8 Northwood. A 3000 winny is a bit slower than both 3.0 C and E's..and a bit faster than a 2.8C


I guess it would be impossible to give them totally accurate PR's but the most glaring example is the 3400. Over 200Pts off what it should be.

The most accuate PR'd chip is the 3000 NC. which incidently 2000 x 1.5 = 3000 :) so it stands to reason.

I like this site because they normalise results over all testing:
http://www.hardware.fr/articles/496/page12.html
 
C

clarkey01

Diamond Member
Feb 4, 2004
3,419
1
0
I believe in my opinion that one of the instigators in AMD?s slumbered PR rating is in Prescott?s bad or little performance increase over Northwood. It would have been great if Prescott was a lot more competitive and maybe AMD had got 90 nm working earlier we could of seen the following:

Assuming, dual channel and 1MB were already implemented like THEY should of done, things may of panned out like this


Clawhammer 3200+ 2Ghz
Clawhammer 3400+ 2.2 Ghz
Clawhammer 3600+ 2.4 Ghz (just for names sake)
Clawhammer 3800+ 2.6 Ghz
Clawhammer 4000+ 2.8 Ghz

AMD would of run into trouble if they gone this rout anyway, presuming Prescott didn?t have the power consumption problems and heat, it would of scaled to 5 Ghz with Tejas in the wings and Nehalem not too far in the distance.

I think AMD have been very lucky with Intels slip up.
 
Zebo

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
presshot was supposed to be slower than northwood clock for clock.. but it was also supposed to be 4.5Ghz by now which would reign superior over any AMD A64 less than 2.8Ghz.. did'nt quite work out as Intel planned.

I think AMD have been very lucky with Intels slip up.
Click to expand...

Not really... People have been predicting for a couple years scaling is dead. AMD took the high road anyway, going for more IPC every iteration of chip while Intel let marketing department guide prosessor development with p4. That's what is paying off. Doing the job right.
 
Zebo

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
Good deal... $100 less than 3800 and just as fast. Course if you overclock any processor over $150 is hard to justify....
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom