ITE IT8620E

ITE IT8620E

I find this above chip on both MBs below, a chip specifically
developed exclusively for Gigabyte.

Gigabyte f(rev6.0) (older)
Gigabyte 78LMT-USB3 R2 (rev1.0) (newer)

Not, long a few months between acquirement and some apparent
adjustments between.

On one of the MBs the ITE IT8620E's six temp sensors, temp 4, reports
operationally near 200F, whereas the other the same chip is closer in
line to low ambient, normal, but no more than 130F (for a couple of
reports).

HWInfo32 V4.38

Also found Gigabyte Pacific Rim response that the sensor relates to
MOSFET and 100C is normal. Author of HWInfo32 looked at it with a
mention he'd address the code for possible revision/update to the ITE
IT8620E.

There's newer than V4.38, (though Microsoft is on the action with
required their updates to run it), just strange from both XP -and
Win7- the same V4.38 is constant at 100C, whereas the other machine
and alternative 78LMT (whichever) MB reports (an exclusive XP/SP3
vidaudio box) no more than 120F are on the same ITE IT8620E sensor
slot (believe it would be exclusive to sensor #4, although they'll no
doubt deviate at times and conditions -- all the hotter slots, #3-#4,
are presently reading 114F).

I'm just as inclined to a case of Gigabyte engineering "wackiness",
temps by others in forums not being a Gigabyte matter regarded for
rock-solid consideration. Despite GB's MOSFET posted "assurances" and
at odds I'm getting between almost, closely identical boards: readings
from the dark side of the moon and the sun's core.

Screw it. Exclusive and custom chipset GB engineering, I doubt
there'll be a precise explanation.

On Mon, 09 Dec 2019 00:36:09 -0500, Flasherly
wrote:


Also came to mind - I'd swapped CPUs between them a few days ago: a
hotter/est quad AMD (4Ghz) for similar upper-range clocked octal,
although both would be near stock 85watts and nothing more exclusive
to AMD power hogs. If I were to re-swap, PIA, it conceivably could be
then the quad affecting;- presuming I didn't actually see to notice
that 100C/HWInfo32 sensor prior for many months with the octal.

HWInfo32 along with a tweak-overclock utility AMD distributed for the
MB, readily seem to do best offhand on this particular series board
for its wider range of sensors, other monitoring programs necessarily
well may not. (Why developers traditionally tend to stick to Intel.)

Flasherly wrote:
On Mon, 09 Dec 2019 00:36:09 -0500, Flasherly
wrote:


Also came to mind - I'd swapped CPUs between them a few days ago: a
hotter/est quad AMD (4Ghz) for similar upper-range clocked octal,
although both would be near stock 85watts and nothing more exclusive
to AMD power hogs. If I were to re-swap, PIA, it conceivably could be
then the quad affecting;- presuming I didn't actually see to notice
that 100C/HWInfo32 sensor prior for many months with the octal.

HWInfo32 along with a tweak-overclock utility AMD distributed for the
MB, readily seem to do best offhand on this particular series board
for its wider range of sensors, other monitoring programs necessarily
well may not. (Why developers traditionally tend to stick to Intel.)

The IT8620E should not be "exclusive" to anybody.

http://www.datasheet-pdf.com/datashe...php?id=1213412

There appear to be three temperature inputs on it. Pin 117,118,119.

https://i.postimg.cc/G2k59tsx/IT8620E.gif

It could be there is a second device which measures "Temp 4".

Try Speedfan and see what devices it detects.

http://www.almico.com/instspeedfan452.exe

Paul

On Mon, 09 Dec 2019 03:47:52 -0500, Paul
wrote:

The IT8620E should not be "exclusive" to anybody.

http://www.datasheet-pdf.com/datashe...php?id=1213412

There appear to be three temperature inputs on it. Pin 117,118,119.

https://i.postimg.cc/G2k59tsx/IT8620E.gif

It could be there is a second device which measures "Temp 4".

Try Speedfan and see what devices it detects.

http://www.almico.com/instspeedfan452.exe

Paul

I read the IT8620E was contracted by or to Gigabyte, either that or I
misread it. HWInfo32 does show, however, six readings to the IT8620E
(multiplexed?).

My BIOS, it occurred to me, everything supportive to AMD's power
throttling I've turned off on this quad now, which I can turn all back
on to boot to Win7 to see if there's a difference. I may also be able
to give SpeedFan a straight shot in its newer version, that is without
encountering any additional Microsoft Win7 updates as was the case
with a newer HWInfo32/64. Offhand I didn't think to expect of
Speedfan IT8620E coverage, despite just earlier looking but dismissing
it, to go back again for a look.

On Mon, 09 Dec 2019 04:13:29 -0500, Flasherly
wrote:
My BIOS, it occurred to me, everything supportive to AMD's power
throttling I've turned off on this quad now, which I can turn all back
on to boot to Win7 to see if there's a difference.

-

Yeah. Big Time. Enabled AMD's "Cool&Quiet" Control along with APM
Master Mode.

HWInfo32 - for CPU speeds (didn't get into voltages or memory
timings), reported the BIOS result as dropping the CPU into a little
less half (AUTO ID) speed , presumably with "OverDrive" as well
enabled to load it, optionally, back up to its rated 4GHz;- I didn't
bother with a full-cores load test to verify.

What I noticed and it didn't take long is the IT8620E dropped by 25%
down into near 150F degrees from an inital Win7 load, slightly less,
of about 185-ish.

Supposedly (from a forum) and Gigabyte reply to someone, that initial
100C or MOSFET/voltage-related heat thingy, which they're calling
normal for the IT8620E reportage, is perhaps a default I'm running.

But I also recall some other shenanigans either along the same
conversation if not another user, reporting they'd also made
adjustments to see perhaps a similar temperature drop, even if lower
than my mid-150F, I can't recall.

Evidently it's a valid Gigabyte thing, I've only picked up with
HWInfo32, very likely, where at the author's forums, I was reading the
above allusions. (AMD's distributed OverDrive (overclocking) is more
the GUI-ooey profiler than oriented to any depth on the boards
sensors).

Which, all in all, makes me wonder that ALL if not more along abreast
MBs currently may be doing this MOSFET thing at 100C, potentially, at
a norm for what is involved with MOSFET are actually doing (voltage
regulations?).

That the IT8620E reports it, as it were, front and center to HWInfo,
at least, might seem a little disconcerting for those not accustomed
to MBs with circuitry normally operative at 100C. Especially with
Gigabyte's semi-pidgin reply, reposted by a user from Pacific Rim GB
facilities, for assertment 100C is A-OK Kewl (in their engineerese,
hardly a sentence to the affirmative the temperature was to be
expected and not of undue concern).

Hey - people *do* start pulling out their IR thermometers and doing
the finger-probe MB grope over less. (What might bug me as much is if
I were to go to all the trouble to swap back in the octal core, I
changed a week or so ago, replacing this quad, to see, say, a drop on
that sensor #4 at 150F or, I suspect, comfortably yet lower (120-135F,
possibly). And then to go over and see what's up with this same MB,
essentially, I've two purchased within a month, no more than two, of
each other, in a slightly different Gigabtye "revision", across the
room).

Flasherly wrote:
On Mon, 09 Dec 2019 04:13:29 -0500, Flasherly
wrote:
My BIOS, it occurred to me, everything supportive to AMD's power
throttling I've turned off on this quad now, which I can turn all back
on to boot to Win7 to see if there's a difference.

-

Yeah. Big Time. Enabled AMD's "Cool&Quiet" Control along with APM
Master Mode.

HWInfo32 - for CPU speeds (didn't get into voltages or memory
timings), reported the BIOS result as dropping the CPU into a little
less half (AUTO ID) speed , presumably with "OverDrive" as well
enabled to load it, optionally, back up to its rated 4GHz;- I didn't
bother with a full-cores load test to verify.

What I noticed and it didn't take long is the IT8620E dropped by 25%
down into near 150F degrees from an inital Win7 load, slightly less,
of about 185-ish.

Supposedly (from a forum) and Gigabyte reply to someone, that initial
100C or MOSFET/voltage-related heat thingy, which they're calling
normal for the IT8620E reportage, is perhaps a default I'm running.

But I also recall some other shenanigans either along the same
conversation if not another user, reporting they'd also made
adjustments to see perhaps a similar temperature drop, even if lower
than my mid-150F, I can't recall.

Evidently it's a valid Gigabyte thing, I've only picked up with
HWInfo32, very likely, where at the author's forums, I was reading the
above allusions. (AMD's distributed OverDrive (overclocking) is more
the GUI-ooey profiler than oriented to any depth on the boards
sensors).

Which, all in all, makes me wonder that ALL if not more along abreast
MBs currently may be doing this MOSFET thing at 100C, potentially, at
a norm for what is involved with MOSFET are actually doing (voltage
regulations?).

That the IT8620E reports it, as it were, front and center to HWInfo,
at least, might seem a little disconcerting for those not accustomed
to MBs with circuitry normally operative at 100C. Especially with
Gigabyte's semi-pidgin reply, reposted by a user from Pacific Rim GB
facilities, for assertment 100C is A-OK Kewl (in their engineerese,
hardly a sentence to the affirmative the temperature was to be
expected and not of undue concern).

Hey - people *do* start pulling out their IR thermometers and doing
the finger-probe MB grope over less. (What might bug me as much is if
I were to go to all the trouble to swap back in the octal core, I
changed a week or so ago, replacing this quad, to see, say, a drop on
that sensor #4 at 150F or, I suspect, comfortably yet lower (120-135F,
possibly). And then to go over and see what's up with this same MB,
essentially, I've two purchased within a month, no more than two, of
each other, in a slightly different Gigabtye "revision", across the
room).

Running the MOSFET at 100C, raises the channel resistance a lot.

And that is a self-defeating design call. You pay good money to
get MOSFETs with low channel resistance, like 35 milliohms.

The only reason they can do that sort of thing,
is the regulator is closed loop and cannot go into
runaway. Modern VRMs have a power limiter, so the
MOSFET cannot self-heat to sky-high values.
(As if 100C isn't already sky-high.)

And I don't believe a temperature readout I see in software,
until I can convince myself the hardware is present to
make those measurements. Maybe Temp4 actually comes out of the
voltage regulator main chip for VCore ? Perhaps the VCore regulator
sits on an I2C bus or something. Maybe they could do something
cheesy, like emulate an LM78 as their temperature measurement
interface. The Speedfan hardware detection log might show
some "strange" entries, which require interpretation.

Paul

On Mon, 09 Dec 2019 07:23:09 -0500, Paul
wrote:

Running the MOSFET at 100C, raises the channel resistance a lot.

And that is a self-defeating design call. You pay good money to
get MOSFETs with low channel resistance, like 35 milliohms.

The only reason they can do that sort of thing,
is the regulator is closed loop and cannot go into
runaway. Modern VRMs have a power limiter, so the
MOSFET cannot self-heat to sky-high values.
(As if 100C isn't already sky-high.)

And I don't believe a temperature readout I see in software,
until I can convince myself the hardware is present to
make those measurements. Maybe Temp4 actually comes out of the
voltage regulator main chip for VCore ? Perhaps the VCore regulator
sits on an I2C bus or something. Maybe they could do something
cheesy, like emulate an LM78 as their temperature measurement
interface. The Speedfan hardware detection log might show
some "strange" entries, which require interpretation.

Paul

Good enough, still not more than $60 for either MB just before they
went out of production (and GB wholly switched into Ryzens for AMD).
Next time I lay this box on its side I'll go through it and touch the
tops of prominent points, any heaksinks, hovering over the rest and
general layout. 100C really is a lot (sensor#4 is reading 185F now,
and as yet another (HWInfo) post mentioned), 'enough to were there
should be a conspicuous "hot spot" of radiant heat' -- something which
occurs to readily detectable by feel alone.

Sure -- Speedfan will be worth a look and I will.

On Mon, 09 Dec 2019 14:45:02 -0500, Flasherly
wrote:

Sure -- Speedfan will be worth a look and I will.

Recall I'd messed with Speedfan after getting these MBs and stuck to
an earlier version, 2005, whereas the latest Speedfan is 2016.

Whatever was bugging me then about Speedfan, I'd switched HWInfo32 as
result: Perhaps more granularity in reporting, maybe considerations
for an event/action HWInfo initiation sequence in the event of an
overtemp condition.

Both Speedfan versions, however, do run. The newest does "see" the
ITE IT8620E:
SuperIO Chip=IT8620E

Configuration is a General heading of IT8620E, three instances of Temp
1, 2, and 3, each expandable for Pwm1-3, then again Pwm 1-5,
respective to SB7xx/8xx PM2 in the first set, and IT8620E for the
ensuing entries 1-5.

Enabling all of them, Temp 1,2, & 3, results in a main menu display of
(some sum or lesser granularity) as respectively each (of) Temp 1,2, &
3 individual entry reporting 99F, 90F, 82F. Bottom-most is CORE 54F
(along HDs 0-4 temps, their temps being contingent on same SuperIO
Chip=IT8620E).

No Lord Hummongous readings from Sensor #4's 190F, at least from
Speedfan's last 4.52 version. Speedfan is relatively realistic, not
spooky. I'd kept and was actually running the old version, which only
detected so far to support my HDD temps.

Take a look at HWInfo32 sometime: It's a trip (configuring it for
relevant or edited display entries -- a nightmare of ticks).

On Mon, 09 Dec 2019 16:10:23 -0500, Flasherly
wrote:

Bottom-most is CORE 54F

--

AMD's switch to reporting an "offset" from general operational temps,
which SpeedFan is reporting.

(Core Temp 1.11 is reporting the same (and allows for some adjustments
to that offset)).