Legit Reviews - FX-4100 Quad Core 3.6GHz Bulldozer Processor Review

[zlejedi]

Last I heard late 3Q early 4Q12 on the plus side, that gives AMD time to move BD to Piledriver and hopefully fix a fair number of issues. Trinity is 'supposed' to improve IPC by ~10-15%. Of course, AMD needs this now, but we'll only get BD B3 (next quarter?).

IIRC AMD only promised 10-15% increase in performance per watt. So it might be 5% IPC and 10% clock speed increase at same TDP.

 

[sm625]

Cant explain why some games almost double with HT though. Even if they used 4 cores, I wouldnt expect HT to double the framerate.

Cache thrashing?

 

[apoppin]

Cache thrashing?

No cache thrashing. Guaranteed.

Drivers.

And Intel evidently does work with Devs in gaming to promote HT. In those 4 games, HT probably acts as 2 cores; i would say the performance increase is probably similar to using a dual-core Phenom II and upgrading to a Quad.

The testing shows that there are games that get a nominal increase with HT and i found that most interesting. i guess next, i will test i7 with HT on and off to see if there is any difference with 4-core HT. i will have to add a few more games into my benching for this.

 

[RussianSensation]

The Q6600 had been available for $266 ever since mid-2007. So now instead of being seven years being stuck with quad-cores it'd be six years (when Haswell comes out). Big whoop.

The complains will obviously go up. Again, by 2013 we'll have been stuck with quad-core CPUs from Intel for non-Enthusiast platforms for six whole years. That's an eternity in the technology world.

In the meantime, they could've released a Phenom II X8 with Llano's 3-4% IPC improvements for K10.5 on the 32nm process and it would've had lower power consumption, lower manufacturing costs (a smaller die based on a previous architecture), higher clock speeds (Phenom II X4 980 can reach 4.3GHz pretty comfortably), and would've been significantly faster than the crap they put out with the FX-8100 series.

Pretty much agreed with everything you said. I had a Q6600 for ~ $300 CDN by August 2007. Rumors of Haswell on 1150 again being limited to quad-core is shocking me. I would have expected Intel to offer us a 6-core Haswell for $300 by 2013. However, since most programs don't use more than 4 threads, it's understandable why Intel is pushing IPC and higher clocks rather than more cores on the mainstream. I mean if Haswell can overclock to 5.7ghz and has a 15% IPC advantage over SB, then that'll be about a 36-42% +/- performance gain over the current SB @ 4.6-4.8ghz. Not bad!

But ya, we prob. won't see a time like Pentium D being replaced by E6300/6400 and then just 1 year later Q6600 for $300 (so basically in that span of 2 years we got 2x the performance from E6400 and then double the cores from Q6600 just 1 year later for $300). That would be the same as Haswell 4500k having 8 cores @ 3.3ghz.....

Oh ya, regarding this review, FX-4100 is a joke. You can get Phenom II X4 965 for $115 on Newegg or the $125 960T that can possibly unlock into the X6.

 

[zlejedi]

I don't think those are specifically programmed for HT. From what i see most of the titles than benefit from HT are titles which benefit from 4 cores.

Probably will be much different with i7 as i's rare to find stuff using >4 cores.

 

[OldTechy]

I wanted to see if the current $110 price
for the FX-4100 was an anomaly or was it
priced correctly for its performance class...

bottom line
After compilation and summary-analysis of 'test data' from published
reviews for the AMD FX-4100, it is my opinion that the current price
represents ~20% discount for it's processor performance class.

how did I come to this conclusion?​

...by analyzing the test-data from these published reviews ...
... guru3d < ....... > legitreviews​

the above test-results were compiled, tabularized and summarized
then a relative value (psuedo-rank) was computed
for the FX-4100 and 5 other of the tested processors:

AMD FX-6100 - FX4 ; ( or AMD FX-8150 - fx8)
AMD A8-3850 - a8
AMD X4-980 - x4
AMD X6-1100T - x6
INTEL 2500k - i5
​

the pseudo-ranks computation will range
from 0(worst) to 10(best) for the the other five(5) processors.
Then similarly, the FX4 test results are used to calculate it's rank,
but its value can be less than 0 if its performance is worse
than ALL of the other five (as you will see).

If you have studied the published review-tests,
my results should not be a suprise.

the final summarized-results of this study
can be visualized by this graph;​

​

which plots each processor's current lowest-cost .vs. it's
overall-performance 'psuedo-rank' from each of the product reviews.

note : the trendlines demonstrate that the FX-4100 pricing
is ~20$ lower than that suggested by the trendlines.

these next 2 images detail & summarize
the 'psuedo-rank' calculations for each test.




finally, for completeness, the following images show
the actual test-data form those reviews in tablular form.



I hope you find this report helpfull ...

O.T.

 

[podspi]

I wouldn't necessarily compare the FX with any FM1 or AM3 CPUs, because part of the upside of buying an AM3+ CPU (for now) is that you might already have the platform, and so that is priced in.

The FM1 CPUs have a GPU baked in, which also adds value.


I'm not refuting what you are saying, just that a strict analysis of a AMD CPUs price/performance comparison might lead to misleading results for the two above reasons.

 

[OldTechy]

thanks for your comments ...

My intent was to present the published
test data in a user-friendly manner.

I believe the average consumer, who might be interested in
the AMD FX-4100, could find the presented results more
useful & informative than the unsummarized data on the reviews.

... the test-data was compiled from already published reviews ...
... it was those testers who selected the CPU's to use in comparison.

Obviously, my method was a simple, brute force analysis & comparison
to demonstrate the correlation between Processor Cost and Tested-Performance' of a very-limited set of processors.

In my sub-set, I (generally) chose the
"premium" model of a 'series" for the most contrast.
exception : the 2500k included because of ownership.

re: "... GPU baked-in .."
... whether the IGP is on CPU or on MoBo, the cost-differentail for comperable CPU-Mobo bundles for current AMD Series platforms would not be relevant to the average FX-4100 consumer. imho ...

I wouldn't necessarily compare the FX with any FM1 or AM3 CPUs, because part of the upside of buying an AM3+ CPU (for now) is that you might already have the platform, and so that is priced in.

The FM1 CPUs have a GPU baked in, which also adds value.

I'm not refuting what you are saying, just that a strict analysis of a AMD CPUs price/performance comparison might lead to misleading results for the two above reasons.