A Quick Hammer Poll

Add Your Comments

A year from now, assume all the following:

Performance:

90nm Athlon64 Hammers get introduced at 2.2GHz (3500+), socket 939 only. The initial ones are generally capable of 2.5GHz overclocked with high-end air and moderate overvoltage, and slowly improve over the course of the first half of 2005 to about 2.8GHz.


130nm Hammers are still around, both in socket 939 and socket 754 versions. They get about 2.5GHz by September/launch date with high-end air/moderate overvoltage and only get slightly better than than by next July.

Semprons get introduced next month. They get about 2.5GHz at launch with high-end air/moderate overvoltage and only get slightly better than than by next July.

Clock-for-clock, the 90nm Athlon64 Hammers do about 3% better than the 130nm socket 939s, 8% better than the socket 754s, and 13% better than the Semprons overall in 32-bit (plus can overclock 200-300MHz more by next July).

Windows for x86-64 is out. Only a few games and some professional apps take real advantage of it by next July, but there are signs many more will do so in the next twelve months after that. Assume those games/apps that are x86-64 show a 10-15% improvement over 32-bit versions.

Assume Intel has nothing that matches or beats AMD’s choices in overall performance during the year.

Pricing:

Scenario 1:

90nm Athlon64 Hammers start off at $500 and drop to around $270 by next July. By then, the cheapest socket 939 130nm Hammer is about $160; cheapest socket 754 is about $140. The cheapest socket 754 Sempron is $90.

Assume the average socket 939 mobo will cost about $130 by next July; the average socket 754 will cost about $90. Assume you can get either AGP or PCI Express versions at the same price.

Socket 939 systems will need two matched sticks of RAM, socket 754 systems will only need one.

1) What do you buy over the course of the year?

a) 90nm Hammer
b) 130nm socket 939 Hammer
c) 130nm socket 754 Hammer
d) 130nm socket 754 Sempron
e) Nothing, or none of the above

Please note that for the purposes of this particular survey, “Nothing, or none of the above” covers any and all reasons for not buying.

Scenario 2:

90mn Athlon64 Hammers start off at $500, and drop to about $200 by next July. By then, the cheapest socket 939 130nm Hammer is about $160; cheapest socket 754 is about $140. The cheapest socket 754 Sempron is $90.

Assume the same mobo costs and memory requirement as in scenario 1.

2) What do you buy over the course of the year?

a) 90nm Hammer
b) 130nm socket 939 Hammer
c) 130nm socket 754 Hammer
d) 130nm socket 754 Sempron
e) Nothing, or none of the above

Scenario 3:

90mn Athlon64 Hammers start off at $500, and drop to about $170 by next July. By then, the cheapest socket 939 130nm Hammer is about $140; cheapest socket 754 is about $120. The cheapest socket 754 Sempron is $90.

Assume the same mobo costs and memory requirement as in scenario 1.

3) What do you buy over the course of the year?

a) 90nm Hammer
b) 130nm socket 939 Hammer
c) 130nm socket 754 Hammer
d) 130nm socket 754 Sempron
e) Nothing, or none of the above

Scenario 4:

90mn Athlon64 Hammers start off at $500, and drop to about $170 by next July. By then, the cheapest socket 939 130nm Hammer is about $140; cheapest socket 754 is about $120. The cheapest socket 754 Sempron is $65.

Assume the same mobo costs and memory requirement as in scenario 1.

4) What do you buy over the course of the year?

a) 90nm Hammer
b) 130nm socket 939 Hammer
c) 130nm socket 754 Hammer
d) 130nm socket 754 Sempron
e) Nothing, or none of the above

Scenario 5:

Assume that socket 754 Semprons get x86-64 enabled at some point in time. Would that change any of your decisions (if you chose Sempron earlier on, say “No”)?

a) No.
b) Yes, it would make me buy a Sempron for $90 rather than a more expensive Hammer.
c) Yes, it would make me buy a Sempron for $90 rather than buy nothing.
d) Yes, it would make me buy a Sempron at $65 rather than a more expensive Hammer.
e) Yes, it would make me buy a Sempron at $65 rather than nothing.

Please send your responses to the special email address listed below (this is a new one):

Ed

Leave a Reply

Your email address will not be published. Required fields are marked *