PDA

View Full Version : Socket A & 754 Sempron caution.


Wes Newell
July 30th 04, 06:54 AM
I don't know if this is well known or not but AMD has used different
criteria in giving the Sempron line of cpu's its PR numbers. So don't
buy a Sempron 3100+ expecting it to be faster than an Athlon 64 3000+.
Because it's nowhere close to as fast. AMD ommitted a bunch of the
benchmarks used for Athlon XP's and Athlon 64 so the Semprons could have
higher PR numbers. The sempron 3100+ won't even keep up with an Athlon 64
2800+ And the Sempron for Socket A 2800+ is actually a tbred B core
running slower (12x166) than the XP 2600+ tbred (12.5x166).

As info only, I crossposted this to several groups.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

Hank Kimball
July 30th 04, 08:27 AM
"Wes Newell" > wrote in message
news:[email protected] .net...
>I don't know if this is well known or not but AMD has used different
> criteria in giving the Sempron line of cpu's its PR numbers. So don't
> buy a Sempron 3100+ expecting it to be faster than an Athlon 64 3000+.
> Because it's nowhere close to as fast. AMD ommitted a bunch of the
> benchmarks used for Athlon XP's and Athlon 64 so the Semprons could have
> higher PR numbers. The sempron 3100+ won't even keep up with an Athlon 64
> 2800+ And the Sempron for Socket A 2800+ is actually a tbred B core
> running slower (12x166) than the XP 2600+ tbred (12.5x166).
>
> As info only, I crossposted this to several groups.
>
> --
> Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
> http://mysite.verizon.net/res0exft/cpu.htm

Thank you. I for one recommend not purchasing a 754 socket CPU at all. I
recommend going straight for the 939 based motherboard and CPU, thus you
preserve your investment for at least 2 years. :o)

Hank Kimball
"I have an important message for you, life or death." .. Ok, what is the
message? "What message?"

July 30th 04, 08:31 AM
On Fri, 30 Jul 2004 05:54:17 GMT, Wes Newell
> wrote:

>I don't know if this is well known or not but AMD has used different
>criteria in giving the Sempron line of cpu's its PR numbers. So don't
>buy a Sempron 3100+ expecting it to be faster than an Athlon 64 3000+.
>Because it's nowhere close to as fast. AMD ommitted a bunch of the
>benchmarks used for Athlon XP's and Athlon 64 so the Semprons could have
>higher PR numbers. The sempron 3100+ won't even keep up with an Athlon 64
>2800+ And the Sempron for Socket A 2800+ is actually a tbred B core
>running slower (12x166) than the XP 2600+ tbred (12.5x166).

>As info only, I crossposted this to several groups.

Once again, your ahead of the game - thanx for the ' heads up ' Wes !

BoroLad

rstlne
July 30th 04, 10:38 AM
"Wes Newell" > wrote in message
news:[email protected] .net...
> I don't know if this is well known or not but AMD has used different
> criteria in giving the Sempron line of cpu's its PR numbers. So don't
> buy a Sempron 3100+ expecting it to be faster than an Athlon 64 3000+.
> Because it's nowhere close to as fast. AMD ommitted a bunch of the
> benchmarks used for Athlon XP's and Athlon 64 so the Semprons could have
> higher PR numbers. The sempron 3100+ won't even keep up with an Athlon 64
> 2800+ And the Sempron for Socket A 2800+ is actually a tbred B core
> running slower (12x166) than the XP 2600+ tbred (12.5x166).
>
> As info only, I crossposted this to several groups.
>


Yea, The benchark is supposed to be a close to a CeleronD comparision
chart..

Derek Baker
July 30th 04, 01:23 PM
Hank Kimball wrote:
> "Wes Newell" > wrote in message
> news:[email protected] .net...
>> I don't know if this is well known or not but AMD has used different
>> criteria in giving the Sempron line of cpu's its PR numbers. So don't
>> buy a Sempron 3100+ expecting it to be faster than an Athlon 64
>> 3000+. Because it's nowhere close to as fast. AMD ommitted a bunch
>> of the benchmarks used for Athlon XP's and Athlon 64 so the Semprons
>> could have higher PR numbers. The sempron 3100+ won't even keep up
>> with an Athlon 64 2800+ And the Sempron for Socket A 2800+ is
>> actually a tbred B core running slower (12x166) than the XP 2600+
>> tbred (12.5x166).
>>
>> As info only, I crossposted this to several groups.
>>
>> --
>> Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
>> http://mysite.verizon.net/res0exft/cpu.htm
>
> Thank you. I for one recommend not purchasing a 754 socket CPU at
> all. I recommend going straight for the 939 based motherboard and
> CPU, thus you preserve your investment for at least 2 years. :o)
>

Looks like I'm just about to do the opposite and get a Socket 754 CPU. Can't
really justify the extra for a 3500+, when I expect to upgrade the CPU and
board together.

--
Derek

Wes Newell
July 30th 04, 01:48 PM
On Fri, 30 Jul 2004 13:32:40 +0000, wrote:

>> Looks like I'm just about to do the opposite and get a Socket 754 CPU.
>> Can't really justify the extra for a 3500+, when I expect to upgrade
>> the CPU and board together.
>
> If your willing to go to a 754 then you might as well just stick with
> your skt A (if that's what you have the 939 kicks ass compared (in all
> of the benchmarks, and from the huge gaps i would tend to belive that it
> would really affect the feeling you get)

You must be looking at different benchmarks than I am. Except for
bandwidth, they seem pretty close, model for model. The only advantage to
939 over 754 is the dual channel ram. And of course future upgrades past
3700+. Starting with a 3000+ that still leaves room. The price difference
is a lot greater than the performance difference at this time.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

Derek Baker
July 30th 04, 01:50 PM
rstlne wrote:
>>> Thank you. I for one recommend not purchasing a 754 socket CPU at
>>> all. I recommend going straight for the 939 based motherboard and
>>> CPU, thus you preserve your investment for at least 2 years. :o)
>>>
>>
>> Looks like I'm just about to do the opposite and get a Socket 754
>> CPU. Can't really justify the extra for a 3500+, when I expect to
>> upgrade the CPU and board together.
>>
>> --
>> Derek
>
>
> If your willing to go to a 754 then you might as well just stick with
> your skt A (if that's what you have)
> the 939 kicks ass compared (in all of the benchmarks, and from the
> huge gaps i would tend to belive that it would really affect the
> feeling you get)

What benchmarks are you looking at? The ones I've seen show a small
difference between the 3400+ and 3500+.

--
Derek

Derek Baker
July 30th 04, 02:15 PM
Derek Baker wrote:
> rstlne wrote:
>>>> Thank you. I for one recommend not purchasing a 754 socket CPU at
>>>> all. I recommend going straight for the 939 based motherboard and
>>>> CPU, thus you preserve your investment for at least 2 years. :o)
>>>>
>>>
>>> Looks like I'm just about to do the opposite and get a Socket 754
>>> CPU. Can't really justify the extra for a 3500+, when I expect to
>>> upgrade the CPU and board together.
>>>
>>> --
>>> Derek
>>
>>
>> If your willing to go to a 754 then you might as well just stick with
>> your skt A (if that's what you have)
>> the 939 kicks ass compared (in all of the benchmarks, and from the
>> huge gaps i would tend to belive that it would really affect the
>> feeling you get)
>
> What benchmarks are you looking at? The ones I've seen show a small
> difference between the 3400+ and 3500+.

Here's a comparison of the 3400+ and 3500+:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2065&p=15

Though note since then the 3400+ ha slost half it's cache and gained 200Mhz.

--
Derek

rstlne
July 30th 04, 02:32 PM
> > Thank you. I for one recommend not purchasing a 754 socket CPU at
> > all. I recommend going straight for the 939 based motherboard and
> > CPU, thus you preserve your investment for at least 2 years. :o)
> >
>
> Looks like I'm just about to do the opposite and get a Socket 754 CPU.
Can't
> really justify the extra for a 3500+, when I expect to upgrade the CPU and
> board together.
>
> --
> Derek


If your willing to go to a 754 then you might as well just stick with your
skt A (if that's what you have)
the 939 kicks ass compared (in all of the benchmarks, and from the huge gaps
i would tend to belive that it would really affect the feeling you get)

Post Replies Here Please
July 30th 04, 02:53 PM
>>>>> "rstlne" == rstlne > writes:

rstlne> If your willing to go to a 754 then you might as well just
rstlne> stick with your skt A (if that's what you have) the 939 kicks
rstlne> ass compared (in all of the benchmarks, and from the huge
rstlne> gaps i would tend to belive that it would really affect the
rstlne> feeling you get)


Really. Anyway the few benchmarks I have seen don't show that
significant of a performance difference between 939 vs 754. Could you
kindly send a url that shows these big differences.

Thanks

Alan

Ed Light
July 30th 04, 08:18 PM
"rstlne" > wrote
> If your willing to go to a 754 then you might as well just stick with your
> skt A (if that's what you have)
> the 939 kicks ass compared (in all of the benchmarks, and from the huge
gaps
> i would tend to belive that it would really affect the feeling you get)

I read that 939 requires matched pairs of memory. Do they have to be
identical, or can they be different brands of the same type?


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

The Chief
July 30th 04, 09:32 PM
The safest thing to do is to purchase a matched set. Same size, same
speed, and same brand. One of the problems is matching up all the
timing. If you look at the specs for different brands, you will see
they use different timings. And it may well end up being the cheapest.
If you don't go with a matched set and you start having weird
problems, like spontaneous rebooting, you will have to then buy a
matched set to fix your unexplained problems.


Ed Light wrote:
> "rstlne" > wrote
>
>>If your willing to go to a 754 then you might as well just stick with your
>>skt A (if that's what you have)
>>the 939 kicks ass compared (in all of the benchmarks, and from the huge
>
> gaps
>
>>i would tend to belive that it would really affect the feeling you get)
>
>
> I read that 939 requires matched pairs of memory. Do they have to be
> identical, or can they be different brands of the same type?
>
>

Wes Newell
July 30th 04, 10:24 PM
On Fri, 30 Jul 2004 07:58:09 -0500, Ed wrote:

> On Fri, 30 Jul 2004 07:31:29 GMT, wrote:
>
>>On Fri, 30 Jul 2004 05:54:17 GMT, Wes Newell
> wrote:
>>
>>>I don't know if this is well known or not but AMD has used different
>>>criteria in giving the Sempron line of cpu's its PR numbers. So don't
>>>buy a Sempron 3100+ expecting it to be faster than an Athlon 64 3000+.
>>>Because it's nowhere close to as fast. AMD ommitted a bunch of the
>>>benchmarks used for Athlon XP's and Athlon 64 so the Semprons could have
>>>higher PR numbers. The sempron 3100+ won't even keep up with an Athlon 64
>>>2800+ And the Sempron for Socket A 2800+ is actually a tbred B core
>>>running slower (12x166) than the XP 2600+ tbred (12.5x166).
>>
>>>As info only, I crossposted this to several groups.
>>
>>Once again, your ahead of the game - thanx for the ' heads up ' Wes !
>>
>>BoroLad
>
> Old news, anyway, who looks at model numbers?
>
Well, my guess is that 90% of the buying public look at no more than the
model numbers. To them, the bigger the better, which isn't the case now.
AMD is basically pulling a smooth marketing ploy. Giving the Sempron
bigger numbers and pricng them higher than competing faster cpu's. This
way it looks like they aren't raising prices, which is exactly what they
are really doing by then discontinuing the XP/Duron line of cpu's.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

Ed Light
July 30th 04, 10:53 PM
I was hoping to use my two pc2700/333 Corsair Value Selects, but the two are
totally different and not actually made by Corsair.

--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Ben Pope
July 30th 04, 11:07 PM
Ed Light wrote:
> I was hoping to use my two pc2700/333 Corsair Value Selects, but the two
> are totally different and not actually made by Corsair.


It should be fine, just set the timings such that both DIMMS are within spec
for all parameters.

Would probably be better to use PC3200 though.

Ben
--
A7N8X FAQ: www.ben.pope.name/a7n8x_faq.html
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...

Ed Light
July 31st 04, 04:08 AM
"Ben Pope" > wrote in message
...
> Ed Light wrote:
> > I was hoping to use my two pc2700/333 Corsair Value Selects, but the two
> > are totally different and not actually made by Corsair.
>
>
> It should be fine, just set the timings such that both DIMMS are within
spec
> for all parameters.
>
> Would probably be better to use PC3200 though.

Yes, it will have to scale it down from the bus, instead of up from 133/266
as it is now (still runs faster than in synch on benchmarks).

In interim, I've just got a 2200+ 35w Barton mobile I'm going to make a
3000+ (2150 @ 333 @ 1.5v) with the wires in the socket method - I hope it
works on my KT400.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Ben Pope
July 31st 04, 09:37 AM
Ed Light wrote:
> Yes, it will have to scale it down from the bus, instead of up from
> 133/266 as it is now (still runs faster than in synch on benchmarks).

It probably will be faster on a VIA.

> In interim, I've just got a 2200+ 35w Barton mobile I'm going to make a
> 3000+ (2150 @ 333 @ 1.5v) with the wires in the socket method - I hope it
> works on my KT400.


Go for it, don't see why not.

Ben
--
A7N8X FAQ: www.ben.pope.name/a7n8x_faq.html
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...

Ben Pope
July 31st 04, 09:48 AM
Ed wrote:
> On Fri, 30 Jul 2004 21:24:27 GMT, Wes Newell
> > wrote:
>
>
>>>
>>> Old news, anyway, who looks at model numbers?
>>>
>> Well, my guess is that 90% of the buying public look at no more than the
>> model numbers. To them, the bigger the better, which isn't the case now.
>
> I meant who in these groups look or cares what the model numbers are? ;p
> Ed

I would look at the model number.

Then I would go find out what differences the core has to the other cores I
know. Then I would guess at how they perform. Then I would also then go
look at some benchmarks.

Ben
--
A7N8X FAQ: www.ben.pope.name/a7n8x_faq.html
Questions by email will likely be ignored, please use the newsgroups.
I'm not just a number. To many, I'm known as a String...

Wes Newell
July 31st 04, 10:44 AM
On Sat, 31 Jul 2004 01:48:55 -0500, Ed wrote:

> On Fri, 30 Jul 2004 21:24:27 GMT, Wes Newell
> > wrote:
>
>
>>>
>>> Old news, anyway, who looks at model numbers?
>>>
>>Well, my guess is that 90% of the buying public look at no more than the
>>model numbers. To them, the bigger the better, which isn't the case now.
>
> I meant who in these groups look or cares what the model numbers are? ;p
>
I think you'd be amazed at the number. I think the people that know what
to look for wouldn't make up more than 25% even here. There's lots of
people that come here simply because they need help.:-)

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

Lord Blue
August 8th 04, 02:58 PM
The 939 does not require matched set of memory, the 940 did/does, but this
is only for older Athlon FX and Opteron systems.

Lord Blue
"Ed Light" > wrote in message
news:[email protected]
>
> "rstlne" > wrote
> > If your willing to go to a 754 then you might as well just stick with
your
> > skt A (if that's what you have)
> > the 939 kicks ass compared (in all of the benchmarks, and from the huge
> gaps
> > i would tend to belive that it would really affect the feeling you get)
>
> I read that 939 requires matched pairs of memory. Do they have to be
> identical, or can they be different brands of the same type?
>
>
> --
> Ed Light
>
> Smiley :-/
> MS Smiley :-\
>
> Send spam to the FTC at
>
> Thanks, robots.
>
>

Wes Newell
August 8th 04, 04:34 PM
On Sun, 08 Aug 2004 16:23:03 +0000, wrote:

>
> "Lord Blue" > wrote in message
> news:[email protected]_s54...
>> The 939 does not require matched set of memory, the 940 did/does, but this
>> is only for older Athlon FX and Opteron systems.
>>
>
> I think it's more to the point that 940 boards require "Registerd" memory..
> I could be wrong but I think that your still limited to running Dual-Chan
> ram only when the pairs are running at the same settings. (Matched settings
> and values) but I could be wrong.

That's totally correct. The ram doesn't have to be "matched" other than
they will work at the same timings. The ram could be from 2 completely
different manufactures. IOW's matched pairs as advertised simply means 2
sticks of the same make/model number /brand.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

rstlne
August 8th 04, 05:23 PM
"Lord Blue" > wrote in message
news:[email protected]_s54...
> The 939 does not require matched set of memory, the 940 did/does, but this
> is only for older Athlon FX and Opteron systems.
>

I think it's more to the point that 940 boards require "Registerd" memory..
I could be wrong but I think that your still limited to running Dual-Chan
ram only when the pairs are running at the same settings. (Matched settings
and values) but I could be wrong.