Trusty 1ghz Xbox owners here?

[N64 freak]

Well we should set up an identical testing environment.
Every Xbox should run the same version of XBMC and we should use the same video files for testing.
So if anybody could just drop in a few different video files that would be cool.

I can't do any test till now as i haven't had the time to solder in the 4 additional Rams.

Will keep you updated on the temps and Frame rates as soon as i got it finished.

[nayparm]

Cheers, I appreciate your time spent on your project and shareing with us. I know how time gets eaten away with these things. It seems like 10 mins have gone and its been about 2 hours lol and there's a moaning female somewhere (although there's always the facebook argument to fall back on )

[Nextelhalo]

I've only tried a 720p movie once on my Trusty machine and it choked badly but I didnt really change any settings when i was converting it. I guess i should get off my lazy ass and read the thread about that so I can report back

[N64 freak]

Just tested some 720p materials on my modded 1ghz xbox and it's no fun.
As NextHalo said they choke badly. I tested a few different compressions but all of them choked bad.
The limiting factor is still the CPU as it allways runs at 100% usage.
Any ideas how to improve that?

I don't mean improve the CPU speed, rather the settings in XBMC.

[spicemuseum]

Just tested some 720p materials on my modded 1ghz xbox and it's no fun.
...The limiting factor is still the CPU as it allways runs at 100% usage.
Any ideas how to improve that?...settings in XBMC.

(Assuming your'e using H.264 (x264) encoded 720p videos) Have you tried setting the H.264 skip loop filter to "skip everything" ?
http://www.xbmc4xbox.org.uk/wiki/UserDa ... ttings.xml
...it's exposed in the settings GUI too under Settings -> System.

[N64 freak]

Just tested it but it won't really change anything.
The Hardware just is to slow for HD decoding in my eyes. But if someone knows a way please proof me wrong!

[Geeba]

H264 is never going to be much good on the xbox... even with 1Ghz & 1.4Ghz CPU's - from memory before the days of hardware acceleration in XBMC minimum spec CPU for HD was a core2 @ 2Ghz, I seem to remember Microsoft saying something similar too....

How does high bitrate AVI play? a stock xbox does ok at HD AVI but I still get a few drops here and there....

[spicemuseum]

Just tested it but it won't really change anything.
The Hardware just is to slow for HD decoding in my eyes.

Yes it is. If your 720p H.264 video is encoded at more than about 1mb/s it will (as a minimum) drop frames on a standard X80X.
Did you try a feature-length film rip of about 600~700MB total file size? They often have serious frame rate drops when scenes get busy, but on fairly static images manage to just about hold on - this marginal case is what I would expect might be just about playable on your 1GHz X80X (especially with the skip loop filter trick - which is what helps iPlayer 1500kb/s streams).

[N64 freak]

Don't have any video material like this around at the moment.
But i will try to get a few good video samples to test out what the upgrade really boosts at HD playback.
But for now i think the only real benefit is for Linux and emulator usage.
Or to even out a few of the smaller performance drops of XBMC.

[r32crazy01]

Going back to the pin modifications, am I correct in thinking you need to connect pins GHI# to VCCT to bypass the SpeedStep?? But what two traces need to be cut?

EDIT: Actually is GHI# connected to VSS (ground)?

Thanks

Intel's Info:

The GHI# signal controls which operating mode bus ratio is selected
in a mobile Pentium III processor featuring Intel SpeedStep
technology. On the processor featuring Intel SpeedStep technology,
this signal is latched when BCLK restarts in Deep Sleep state and
determines which of two bus ratios is selected for operation. This
signal is ignored when the processor is not in the Deep Sleep state.
This signal is a "Don't Care" on processors that do not feature Intel
SpeedStep technology. This signal has an on-die pull-up to VccT and
should be driven with an Open-drain driver with no external pull-up

[Floydthebarber]

Okay, the long story is just about technical research.
After getting frustrated with the 933mhz Problem i dropped the work on it.
I got help from the user NextHalo who has shown me that his xbox has exactly the same visible mod's.
So i was sure with the Modell of the CPU and so on but i missed some modification that you can't see.
just a hint, they are hidden under the CPU!
I had to cut 2 traces and set one connection between two pins. The xbox now reads 1000mhz.

So summarized i just missed to disable speedstep and get the xbox to read the FSB correct.

Hey, how's it going? So it looks like you actually just desoldered the bga package off of the SL53S, and left off bottom part with the pins. Am I correct on this? Can you please let me know which two traces you cut and the connection you made between two pins? The help would be greatly appreciated. I could pay you for this information.

[professor_jonny]

the two pins are tied to vcc with pullups internally in the processor core forcing it to startup in a low power state by default even if the if the core voltage is at running at 1.6v.

putting the bga rework stations for the xbox360's to good use i see

[r32crazy01]

the two pins are tied to vcc with pullups internally in the processor core forcing it to startup in a low power state by default even if the if the core voltage is at running at 1.6v.

Professor Johnny is VSS the low signal? Would the below info mean join VSS and GHI# for “Maximum performance” mode?

Thank you

An operating mode transition to the high core frequency can be made by putting the processor into the
Deep Sleep state, raising the core voltage, setting GHI# low, and returning to the Normal state.
This puts the processor into the “Maximum performance” mode. A transitioning back to the low-
core frequency can be made by reversing these steps.

In this document a “#” symbol following a signal name indicates that the signal is active low. This
means that when the signal is asserted (based on the name of the signal) it is in an electrical low
state.

[professor_jonny]

I can't download the 25mb developer doc's of the intel site from home but i found some info but it looks like you drive it logic low.



various people have had this problem on specific laptops because the speed step control was not imbedded in to the motherboard and caused it to perminantly run in the low power state.

Enhanced Intel SpeedStep technology allows the processor to switch between two core frequencies
automatically based on CPU demand, without having to reset the processor or change the system
bus frequency. The processor has two bus ratios and voltages programmed into it instead of one
and the GHI# signal controls which bus ratio and voltage is used. After reset, the processor will
start in the lower of its two core frequencies, the ?Battery Optimized? mode. An operating mode
transition to the high core frequency can be made by setting GHI# low, putting the processor into
the Deep Sleep state, regulating to the new VID output, and returning to the Normal state. This puts
the processor into the high core frequency, or ?Maximum Performance? operating mode. Going
through these steps with GHI# set high, transitions the processor back to the low core frequency
operating mode. The processor will drive the VID[4:0] pins with the VID of the current operating
mode and the system logic is required to regulate the core voltage within specification for the
driven VID.

The mobile Pentium III processor featuring Intel SpeedStep technology is specified to operate in
either of two modes, the “Maximum Performance Mode” or the “Battery Optimized Mode”.

frequency and voltage pair identifies the operating mode. The voltage provided to the processor
must meet the core voltage specification for the current operating mode. If an operating mode
transition is made, then the system logic must direct the voltage regulator to regulate to the voltage
specification of the other mode. After reset, the processor will start in the lower of its two core
frequencies, so the core voltage must meet the lower voltage specification. Any RESET# assertion
will force the processor to the lower frequency, and the core voltage must behave appropriately.
INIT# assertions ("soft" resets) and APIC bus INIT messages do not change the operating mode of
the processor. Some electrical and thermal specifications are for a specific voltage and frequency.
The mobile Pentium III processor featuring Intel SpeedStep technology will meet the electrical and
thermal specifications specific to the current operating mode and is not guaranteed to meet the
electrical and thermal specifications specific to the opposite operating mode. The timing
specifications in Table 22 must be met when performing an operating mode transition.

these are the pins to look into their function:

and here is a data sheet referenced for speed step in the p3 chip
http://download.intel.com/design/mobile ... /p3_ds.pdf


CLKREF System bus clock trip point control
CMOSREF 1.5V CMOS input buffer trip point control
EDGECTRLP GTL+ output buffer control
GHI# Intel SpeedStep technology operating mode selection
BSEL[1:0] Processor system bus speed selection
RTTIMPEDP On-die GTL+ termination control
VCCT On-die GTL+ termination current supply
VID[4:0] Voltage Identification

[r32crazy01]

Thanks for the info Professor Johnny

I'm a newb to some terms used but this https://learn.sparkfun.com/tutorials/logic-levels/all article helped me.

EDIT:
From my research on the Intel docs it looks like the steps are as follows?
1.) GHI# Join to VSS
2.) Cut trace BCLK#
3.) Cut trace STPCLK#

[professor_jonny]

Edit: looks like the trace needs to be cut to BCLK as well(puts it into DeepSleep) otherwise GHI signal is ignored.

looking at the timing charts i think it says it ignores that pin for 10us on power up to enable it to start direct in high power state with out forcing deep sleep.

[Geeba]

Still curious on the bus speed of the 1Ghz CPU? is it 133mhz the same as the 733 CPU? I seem to remember the 1.5Ghz DreamX loosing out on power due to its 100Mhz bus...

I'm guessing the CPU denotes the speed of the clock so there is no way of overclocking?

Also found these:-

http://www.fuzing.com/vli/0000043bb223/ ... de-Adapter

There's a few other options at the bottom....

[r32crazy01]

Still curious on the bus speed of the 1Ghz CPU? is it 133mhz the same as the 733 CPU? I seem to remember the 1.5Ghz DreamX loosing out on power due to its 100Mhz bus...

I'm guessing the CPU denotes the speed of the clock so there is no way of overclocking?

Also found these:-

http://www.fuzing.com/vli/0000043bb223/ ... de-Adapter

There's a few other options at the bottom....

.

Nice find. but unfortunately that adapter is unsuitable as its for a 423 socket (xbox is 495).

[Geeba]

There's a few options at the bottom of that link.... PIII adapters for Tualatin class CPU's etc

[professor_jonny]

the dreamx used a celeron 1.5 with a 100mhz fsb the 1.4ghz has Tualatin core at 160mhz.

on the xbmc forum there was test showing the difference in proformance between the 1.5 and 1.4.
The 1.4 seemed to proform better with the higher bus freq and bigger l2 cache as you would expect but the celeron did excell in some aspects.

if you could find a engineering sample pentium chip they pop up on ebay from time to time you could change the multiplier giving it a bit of extra zing.