FX-4300 FX-6300 FX-8320 Speed Differences

[Damaeus]

I'm curious about why there's the AMD FX-8320 Vishera with an Eight-Core
running 3.5 GHZ ($180), while there's also the AMD FX-4300 Vishera with a
Quad-Core running 3.8 GHz. Why is the Quad-Core rated at a higher clock
speed than the more expensive Eight-Core, not to mention the Six-Core
FX-6300, which is also rated at 3.5 GHz?

The FX-8320 might be more than I want to spend, but I could get the
FX-6300; it's only $10 more than the FX-4300 on newegg.com. ...though the
FX-4300 was only about $79 or $89 a couple of weeks ago. I need choose
something soon before it goes up in price again.

Is a Six-Core at 3.5 GHz going to be faster, overall, than a Quad-Core
running 3.8 GHz?

Damaeus

[Wes Newell]

On Thu, 27 Dec 2012 18:42:37 -0600, Damaeus wrote:

> Is a Six-Core at 3.5 GHz going to be faster, overall, than a Quad-Core
> running 3.8 GHz?

Depends on the threads running. If all 6 cores can be used to the max
then the effective speed is 3.5x6, or 21Ghz, while only 4 is 4x3.8, or
15.2Ghz. So the 6 core is potentially 30% faster.

[Damaeus]

In news:alt.comp.hardware.amd.x86-64, Wes Newell
<w.ne...@invalid.invalid> posted on Fri, 28 Dec 2012 03:43:47 +0000 (UTC)
the following:

Okay, thanks. That makes sense. Learning all the time. :)

I was using a dual-core before I built this temporary system that only has
one core. It's extremely painful to use.

I'm going with the 6-core. The 8-core is a little out of my reach with
the other stuff I need.

Damaeus

[Jim Beard]

For an ordinary desktop user, there is a great advantage in
shifting to a dual-core cpu from single-core. Beyond dual-core,
the benefit of more cores depends on how many processes are
running and how long they need to run at a time for high
throughput If you have a lot of processes going and several that
need a core for long periods of time at a stretch (relatively, of
course, in cpu time -- not long by user time), more cores will do
better. But generally, gains beyond dual-core are usually not
apparent and the added overhead can in fact slow things down.

I have a couple of machines, one dual-core and one eight-core,
both running the same OS, very similar configuration, and similar
workload. The dual-core is actually much snappier in response
time most of the time, but when it gets overloaded (copying
massive files while browsing the 'Net or other multiple heavy
loads) it will bog down. The eight-core is slightly slower but
smoother in response time and never ever gets bogged down enough
to affect more than the one process that is hogging a core.

There is also the consideration that four and six core cpus often
were manufactured as eight-core, and cores that failed QA tests
were disabled. If some of the cores failed, it can suggest a
problem with the entire piece of silicon. That usually shows up
as an infant death, though, so at worst you might need to go
through the return-merchadise-authorization routine to get a
replacement.

Cheers!

jim b.


--
UNIX is not user unfriendly; it merely
expects users to be computer-friendly.

[Yousuf Khan]

On 27/12/2012 7:42 PM, Damaeus wrote:
> I'm curious about why there's the AMD FX-8320 Vishera with an Eight-Core
> running 3.5 GHZ ($180), while there's also the AMD FX-4300 Vishera with a
> Quad-Core running 3.8 GHz. Why is the Quad-Core rated at a higher clock
> speed than the more expensive Eight-Core, not to mention the Six-Core
> FX-6300, which is also rated at 3.5 GHz?

Well, the more cores there are, the more heat the chip overall
generates. They attempt to stay within a certain power consumption
window with these chips, for example 95W. A chip with only 4 cores will
be consuming less power overall, therefore its clock can be cranked up a
bit more and it can still stay within the 95W power limits, than a chip
with 6 or 8 cores.

Now as for the 6 & 8 core chips having the same base frequency of
3.5GHz, I'm willing to bet that the 6-core has a slightly higher turbo
frequency than the 8-core. The turbo frequencies are a temporarily
higher frequency that can be applied to one or two cores at a time as
the need arises, so those cores would be running faster than the base
frequency. Without looking it up, I'm betting that the turbo frequency
of the 8 core would be 4.0GHz, while the turbo frequency of the 6 core
would be about 200Mhz more at 4.2GHz. That's again because a 6 core
would be generating less heat than the 8 core, so it can have a higher
turbo speed.

> The FX-8320 might be more than I want to spend, but I could get the
> FX-6300; it's only $10 more than the FX-4300 on newegg.com. ...though the
> FX-4300 was only about $79 or $89 a couple of weeks ago. I need choose
> something soon before it goes up in price again.
>
> Is a Six-Core at 3.5 GHz going to be faster, overall, than a Quad-Core
> running 3.8 GHz?

It depends on your workload. If you have a lot of programs running in
the background, then the more cores there are, the better. If on the
other hand, you just want your foreground program to run at its fastest
possible speed, then it's likely that a lesser number of cores with
faster cores might be the better way to go.

Yousuf Khan

[Damaeus]

In news:alt.comp.hardware.amd.x86-64, Jim Beard <jdb...@patriot.net>
posted on Fri, 28 Dec 2012 10:54:19 -0500 the following:

> There is also the consideration that four and six core cpus often were
> manufactured as eight-core, and cores that failed QA tests were
> disabled. If some of the cores failed, it can suggest a problem with
> the entire piece of silicon. That usually shows up as an infant
> death, though, so at worst you might need to go through the
> return-merchadise-authorization routine to get a replacement.

Great Scott!!!! Why do they DO these things to us???? "Let's just sell
this defective chip and hope we don't get it back."

I could just throw up and eat it. :(

I liked my dual-core processor. I assumed a six-core would be even
better. I like having more than enough. Even if all I wanted to do is
read e-mail and play Solitaire, I'd want something capable of running all
the latest games. But I DO want to play some of the latest games, so I
got this stuff. Oh man... I had no idea of the sneaky, underhanded
tactics that are being used to sell rejects as lower models to ignorant
consumers. No wonder so many computer parts have about 25% of the people
saying they suck.

Damaeus

[Damaeus]

In news:alt.comp.hardware.amd.x86-64, Yousuf Khan
<bbb...@spammenot.yahoo.com> posted on Sat, 29 Dec 2012 01:03:08 -0500 the
following:

> Well, the more cores there are, the more heat the chip overall
> generates. They attempt to stay within a certain power consumption
> window with these chips, for example 95W. A chip with only 4 cores will
> be consuming less power overall, therefore its clock can be cranked up a
> bit more and it can still stay within the 95W power limits, than a chip
> with 6 or 8 cores.
>
> Now as for the 6 & 8 core chips having the same base frequency of
> 3.5GHz, I'm willing to bet that the 6-core has a slightly higher turbo
> frequency than the 8-core.

The six-core turbo speed is 100 MHz faster than one of the eight-core
turbo speeds, but the other eight-core turbo speed is 100MHz faster than
the six-core turbo speed. The list is below:

Here's a list of the four Vishera AM3+ CPUs from the Newegg.com website:

AMD FX-8350 Vishera 4.0GHz (4.2GHz Turbo) Socket AM3+ 125W Eight-Core
AMD FX-8320 Vishera 3.5GHz (4.0GHz Turbo) Socket AM3+ 125W Eight-Core
AMD FX-6300 Vishera 3.5GHz (4.1GHz Turbo) Socket AM3+ 95W Six-Core
AMD FX-4300 Vishera 3.8GHz (4.0GHz [Turbo]) Socket AM3+ 95W Quad-Core

> The turbo frequencies are a temporarily higher frequency that can be
> applied to one or two cores at a time as the need arises, so those
> cores would be running faster than the base frequency. Without looking
> it up, I'm betting that the turbo frequency of the 8 core would be
> 4.0GHz, while the turbo frequency of the 6 core would be about 200Mhz
> more at 4.2GHz. That's again because a 6 core would be generating less
> heat than the 8 core, so it can have a higher turbo speed.

As long as the factory heatsink and the heat transfer material they apply
to it works, I'm happy. I have a little vial of Arctic Silver, however,
that I used on my still-older computer, an Abit KT7A-RAID machine. But
once I put a heatsink on, I don't like to take it off unless I just have
to. If a six-core processor really is a defective eight-core processor,
maybe they shipped it with a heatsink and thermal compound that's also
suited for an eight-core processor. My dual-core machine had no heat
issues with the CPU whatsoever, and I just left the factory thermal pad do
what it was designed to do. Since I won't be overclocking, I'm planning
to do the same with the new build.

> Damaeus posted:
>
> > Is a Six-Core at 3.5 Ghz going to be faster, overall, than a Quad-Core
> > running 3.8 Ghz?

>
> It depends on your workload. If you have a lot of programs running in
> the background, then the more cores there are, the better. If on the
> other hand, you just want your foreground program to run at its fastest
> possible speed, then it's likely that a lesser number of cores with
> faster cores might be the better way to go.

I see the point now. I don't typically do a LOT of multitasking, but
sometimes I do. I do play some Facebook games, and those are amazingly
processor-intensive. I know it ran my dual-core processor fluctuating
between 35% and 60%, even more than Final Fantasy XI, which is a real,
full-screen video game.

As for Facebook games, sometimes I like to have a couple or three games
running in different tabs in the browser. I imagine more cores will come
in handy just for that alone. This single-core processor will barely run
one game at a time, and it doesn't do that well. What flows like
chocolate syrup on my friend's computer looks like a stuttering annoyance
on this machine. My old dual-core ran the same games even more fluidly
than my friend's computer, and they're both dual-cores. Mine had a
dedicated GeForce 7950GTOC, while his is an integrated Intel G31/33
integrated in the motherboard's chipset. His won't even display Farmville
2 properly. Many of the Farmville 2 game elements are completely missing
on his, both static elements and animations.

On this single-core machine, there's one game that even causes my mouse to
intermittently hang just because of a Flash advertisement running on the
same web page.

Still, just the fact that each core runs at 3.5GHz core is still going to
be better than the 2.2 GHz dual-core it's really replacing, not counting
this dreadful single-core 1.83 GHz one. If I had it to do over again, I
might have just gone with the quad-core after all I've learned. But I
figured that since the six-core was only ten dollars more, why not go for
it. Hopefully it'll just work as it's supposed to without giving me any
problems. I'm sure it'll be generally faster and I will still be happy
with it since it'll be faster than the rig that died, and much faster than
this piece of garbage I'm using now. This would make a fine dedicated
word processor and web browser, but it's not even fit for the simplest
browser games.

Damaeus

[Jim Beard]

On 12/30/2012 05:33 AM, Damaeus wrote:
> In news:alt.comp.hardware.amd.x86-64, Jim Beard <jdb...@patriot.net>
> posted on Fri, 28 Dec 2012 10:54:19 -0500 the following:
>
>> There is also the consideration that four and six core cpus often were
>> manufactured as eight-core, and cores that failed QA tests were
>> disabled. If some of the cores failed, it can suggest a problem with
>> the entire piece of silicon. That usually shows up as an infant
>> death, though, so at worst you might need to go through the
>> return-merchadise-authorization routine to get a replacement.
>
> Great Scott!!!! Why do they DO these things to us???? "Let's just sell
> this defective chip and hope we don't get it back."
>
> I could just throw up and eat it. :(

No need to do that.

If you want a chip that passed all QA tests, pay full price for
the 8-core cpu.

If you decide saving money is important to you, you can buy a cpu
that has 4 or 6 cores that passed QA tests, and one or more that
did not. The ones that did not have been disabled, and do
nothing more than take up room on the chip.

There is a slightly higher risk of infant death for a cpu known
to have one or more defects (there could be another, that QA did
not catch), but there is a risk of infant death for all cpus.

Why scrap a cpu with less than 8 cores that passed but that
likely will provide good service for its entire design
service-life? Why not sell it at a discount, and let both
manufacturer and buyer benefit (the manufacturer gets something
though not full price, and the buyer gets a good chip that will
be replaced if it goes bad -- cheap).

[Yousuf Khan]

On 30/12/2012 6:02 AM, Damaeus wrote:
> In news:alt.comp.hardware.amd.x86-64, Yousuf Khan
> <bbb...@spammenot.yahoo.com> posted on Sat, 29 Dec 2012 01:03:08 -0500 the
> following:

>> Now as for the 6& 8 core chips having the same base frequency of

>> 3.5GHz, I'm willing to bet that the 6-core has a slightly higher turbo
>> frequency than the 8-core.
>
> The six-core turbo speed is 100 MHz faster than one of the eight-core
> turbo speeds, but the other eight-core turbo speed is 100MHz faster than
> the six-core turbo speed. The list is below:
>
> Here's a list of the four Vishera AM3+ CPUs from the Newegg.com website:
>
> AMD FX-8350 Vishera 4.0GHz (4.2GHz Turbo) Socket AM3+ 125W Eight-Core
> AMD FX-8320 Vishera 3.5GHz (4.0GHz Turbo) Socket AM3+ 125W Eight-Core
> AMD FX-6300 Vishera 3.5GHz (4.1GHz Turbo) Socket AM3+ 95W Six-Core
> AMD FX-4300 Vishera 3.8GHz (4.0GHz [Turbo]) Socket AM3+ 95W Quad-Core

In the case of the 8320 vs. 8350, that's usually down to a result of
"bin-splitting" which is in a tray full of otherwise identical
processors, some of them will just naturally be able to run at higher
frequencies. The factory tests each chip at certain speeds and sorts
them according to their maximum reliable speeds.

> As long as the factory heatsink and the heat transfer material they apply
> to it works, I'm happy. I have a little vial of Arctic Silver, however,
> that I used on my still-older computer, an Abit KT7A-RAID machine. But
> once I put a heatsink on, I don't like to take it off unless I just have
> to. If a six-core processor really is a defective eight-core processor,
> maybe they shipped it with a heatsink and thermal compound that's also
> suited for an eight-core processor. My dual-core machine had no heat
> issues with the CPU whatsoever, and I just left the factory thermal pad do
> what it was designed to do. Since I won't be overclocking, I'm planning
> to do the same with the new build.

For the most part you should just let the factory thermal pad do its
work. You'll only need the Arctic Silver if you plan to replace the
factory heat sink with an aftermarket one.

>>> Is a Six-Core at 3.5 Ghz going to be faster, overall, than a Quad-Core
>>> running 3.8 Ghz?
>>
>> It depends on your workload. If you have a lot of programs running in
>> the background, then the more cores there are, the better. If on the
>> other hand, you just want your foreground program to run at its fastest
>> possible speed, then it's likely that a lesser number of cores with
>> faster cores might be the better way to go.
>
> I see the point now. I don't typically do a LOT of multitasking, but
> sometimes I do. I do play some Facebook games, and those are amazingly
> processor-intensive. I know it ran my dual-core processor fluctuating
> between 35% and 60%, even more than Final Fantasy XI, which is a real,
> full-screen video game.

I don't think any of the facebook apps will be much of a challenge to
any one of these cores. A world of difference between the old Athlon 64
X2 and these FX chips.


Yousuf Khan

[Damaeus]

In news:alt.comp.hardware.amd.x86-64, Jim Beard <jdb...@patriot.net>

posted on Sun, 30 Dec 2012 11:00:08 -0500 the following:

> If you decide saving money is important to you,

I wish I didn't have to worry about money. If I had more to spend, I
wouldn't care about saving it. But since I only had $827 to spend on new
parts, I had to skimp on cores in case I have to get another hard drive.
And, of course, I needed the motherboard and memory, plus I had to get a
new SATA DVD drive/burner since all I have is the a couple of the old IDE
types. And then there was Windows 8, full version OEM. I had two
identical Western Digital 320GB SATA hard drives, but one went bad within
two years. I'm still using the other one. In the rig I'm replacing, I
was running only that one drive with two partitions. Now that I'm going
to try to run Win8 64-bit and WinXP 32-bit on different partitions, I
might need a second drive, and I'm going to try a solid-state drive if I
actually need to get one. Perhaps I can get a 120GB SSD and give 80GB to
Windows 8 and 40GB to Windows XP, since hopefully all I'll have on WinXP
is Final Fantasy XI. Then I'll have the full 320GB drive for data. But
then, I also need to replace my video capture board. I can do that for
$42. And it would be nice to have front-panel USB ports with a MicroSD
card-reader. I had very little luck finding a front-panel unit that has
card readers and more than one USB port. Most had one USB port. One had
three, but the reviews were horrible, and one reviewer said that his
MicroSD card went into the slot so far that you'd definitely need
fingernails to get it out. My friend's PC has a MicroSD reader and the
card still sticks out about 5/8 of an inch when it's recognized by his
computer.

> you can buy a cpu that has 4 or 6 cores that passed QA tests, and one
> or more that did not. The ones that did not have been disabled, and
> do nothing more than take up room on the chip.

I just assumed, incorrectly, apparently, that a six-core chip had six
cores and they all passed with flying colors.

[...]

> Why scrap a cpu with less than 8 cores that passed but that
> likely will provide good service for its entire design
> service-life? Why not sell it at a discount, and let both
> manufacturer and buyer benefit (the manufacturer gets something
> though not full price, and the buyer gets a good chip that will
> be replaced if it goes bad -- cheap).

Yes, that would be nice. I'll save my receipts, as usual.

Damaeus

[Damaeus]

In news:alt.comp.hardware.amd.x86-64, Yousuf Khan

<bbb...@spammenot.yahoo.com> posted on Sun, 30 Dec 2012 12:53:45 -0500 the
following:

> On 30/12/2012 6:02 AM, Damaeus wrote:

[...]

> > I see the point now. I don't typically do a LOT of multitasking, but
> > sometimes I do. I do play some Facebook games, and those are amazingly
> > processor-intensive. I know it ran my dual-core processor fluctuating
> > between 35% and 60%, even more than Final Fantasy XI, which is a real,
> > full-screen video game.
>
> I don't think any of the facebook apps will be much of a challenge to
> any one of these cores. A world of difference between the old Athlon 64
> X2 and these FX chips.

I hope so. They weren't really a challenge (judging by the smoothness of
playing) even to my Athlon 64 X2 4200+ as long as I only had one game
running at a time, although I think my bad capacitor next to the PCIe slot
was causing some animation problems in certain games.

And I hope I didn't mess things up by buying over-rated parts this time
around. All these discussions in this group and others have got me
worried pretty deeply. Yet when I built my Fatal1ty AN9 32x system, I
didn't think about the speed at which the CPU could utilize memory. I
just got some memory I knew was compatible along with a mid-range
processor, and that was the most stable system I have ever built. Maybe
by happenstance, I just picked out the right hardware without thinking
about it.

I put a lot of thought into what I was doing with this new hardware, but
now I'm all stressed out. I don't care if it's not the fastest computer
in town as long as it's stable. And I always take my time putting it all
together and making sure all the cables and wires are plugged into the
right spots. If I see sparks and smoke, I'll be stuck using this Abit
NF-7 I'm using now, and I think of jokes about 386 machines. With the
processing power that's needed today for certain things, trying to cram
that down this PCs throat does make it act like a 386.

Damaeus