*HP 2510p Owners Lounge*

[Jay2k1]

Here's the PLL chip. For the southbridge, I'd have to open the whole chassis, I wouldn't wanna do that unless absolutely necessary.

So SetFSB does not (yet) support that chip

[wesselcolsen]

Has anybody actually tried attaching a sata drive in the optical drive caddy yet? Does it need a pata to sata converter as well? The caddy I have bought is a sata caddy (not delivered yet) and it would seem like the way to go as I have no real use for the optical drive.

Hope it works!

[nando4]

UPDATE (10-8-10): can flash 2510P-F0F-dualIDA to gain dual_IDA's extra multiplier for a free extra 10-12.5% CPU performance. More gains can be had if combined with setfsb's PLL overclocking.

Consider to the 200Mhz_bclk mod that improves X3100+RAM performance

Suggest use the 2510P Performance Toolkit to easily setup a AC/overclocked and DC/undervolted setup, maintaining settings when doing a resume from standby/hibernate.

INFO: U7600 OC Win7/64 WEI and wprime32 scores

CPU speed FSB Mhz multiplier^1	1.2 133 x9	1.2 200 x6-BC	1.33 133 x10-DI	1.44 160 x9	1.44 240 x6-BC	1.55 194 x8	1.6 160 x10-DI	1.6^3 178 x9	1.6 178 x9-SI	1.66 [unstable] 185 x9-SI
Win7 WEI Processor Memory Graphics^2 Gaming Graph Primary HDD	pic 3.9 4.2 2.8 3.0 5.9*2	pic 3.9 4.6 3.2 3.1 5.9*2	pic 4.1 4.2 2.6 3.0 4.4*1	pic 4.3 4.5 3.1 3.1 5.9*2	pic 4.3 4.9 3.5 3.2 5.9*2	pic 4.4 4.7 3.4 3.2 5.9*2	pic 4.6 4.6 3.1 3.1 6.4*3	pic 4.6 4.6 3.3 3.2 6.4*3	pic 4.8 4.8 3.3 3.2 6.4*3	pic 4.8 4.8 3.4/6.7*4 3.2/6.7*4 5.9*2
wprime 2.00	72.68		65.68	60.40		56.32	54.41	53.62		51.98

Yellow - recommended overclock. Can run 160@x9 (1.44Ghz) on DC at lowest VID. So no need to downclock for optimum battery life.
Magenta - FSB greater than 178Mhz required CAS=5 written to RAM. System unstable at > 195Mhz FSB.
* Modifications: 1=(none)1.8" 80GB ZIF HDD [MK8025GAL], a single-platter HP replacement of the original supplied MK8009GAL. The dual-platter MK8009GAL got only a 3.9 WEI!! 2=2.5" 500GB 5400rpm sata HDD, 3=128GB K3VLAR SSD, 4=HD4670 gpu

^1 - Requires Throttlestop. DI=dualIDA overclock gives extra multiplier. SI=singleIDA overclocks one core with an extra multiplier. BC=200Mhz_bclk modded
^2 - May need Modded X3100 driver for Win7/32. It increases desktop graphics score by ~10%/0.2 points, described here
^3 - losdrivare duplicated these results with the PATA 160GB HM160HC obtaining a 5.5 disk score here

Tested unit: u7600-1.2Ghz 2GB 80GB/4200rpm 2510P. Compare SL9400 2530P, SU9400 R600 WEI here. u7600@1.6Ghz is faster than a SU7300-1.3 Acer 1810T and SU9400-1.4 Lenovo X301 in all areas except 3D graphics, where dual-channel X4500 is faster. If overclocking, no need to pay a premium in s/h market for the x10 multiplier enabled u7700 cpu since it's likely to have the same ~1.66Ghz overclock stability wall as the u7600. The dualIDA modded bios means a U7500-1.06 can likely get a x9 multiplier overclock.

INFO: Overclocking

WARNING: FSB Overclocking means running components faster with an increase in operating temperature. Observe decent cooling precautions. I take no responsibility for any damage. User beware.

setfsb now support 2510P overclocking

Abo has added the 2510P's ICS 9LPRS355BGLF PLL to his setfsb software. Below is a screenshot with a 5% overclock of the FSB and PCI-E clocks on a 1.2Ghz u7600 2510P. Note the bug where the setfsb command has no effect in increasing the FSB after the 2510P has been in a prolonged standby or hibernate.

1: 2510P with 5% FSB and pci-e overclock 2: Error notification when PLL is in a frozen state after a prolonged hibernate or standby. A quick standby followed by resume will unfreeze the PLL and allow the FSB/pci-e clocks to be set *without* the need for a shutdown.

setfsb opens up a lot more performance tweak options. The u7600 is overclocked from 1.2Ghz to 1.4Ghz here. 1.5Ghz is a very easy overclock with the standard voltages and RAM normal 266Mhz memory timings as set during bootup. Setting CAS=5 ram timings provides an extra level stability. The HP supplied RAM is DDR2-667 so it does support 333Mhz.

Overclocking with Linux

See INFO: Overclocking the 2510P in Linux for detailed instructions. Another way is to boot to WinXP/Win7 running setfsb to overclock then upon rebooted (not shutdown) into Linux, the overclocked PLL settings remain. 'dmesg' shows per cpu bogomips@1.2Ghz=2394, bogomips@1.5Ghz=3001 (overclocked). Gnome's CPU Frequency monitor incorrectly shows the speed at 1.2Ghz.

CPU temperature

If overclocking, consider observing CPU temps to ensure operation stays as far as possible below Tjmax of 100C using tools like Everest or rmclock. Better removal of heat can be achieved by application of some decent Artic Silver thermal paste between the CPU/northbridge and the heatsink or clearning/blowing out any dust or gunk accumulation blocking the little fan under the system. Also the fan/airway is easy to block with your knee if sitting on your lap since it's in the back right corner.

Memset, spdtool and thaipoon burner - alter RAM timings

Memset cannot alter the primary timing, CAS latency, but Thaiphoon burner or spdtool can by writing timing data to the RAM's eeprom. If overclocking in 1.55-1.65 Ghz range, then consider modding RAM eeprom to store CAS latency=5 in the 266Mhz timing table. This is the latency used by RAM operating at 333Mhz. Detailed instructions for power users are here. NOTE: Observe tools' warnings to prevent RAM damage.

Hardware Overclocking (Advanced)

The 9LPRS355.pdf datasheet shows that a logic of 1 on the PLL FSLB pin would change the u7xxx CPU FSB to 166Mhz.

FSLC FSLB FSLA CPUFREQ
.....0......0......1.....133.33 <---- default (u7xxx CPU)
.....0......1......1.....166.66 <-----overclock (u7xxx CPU)
.....0......1......0.....200.00 <-----default (L7xxx CPU)
.....0......0......0.....266.66 <-----b-i-g overclock (L7xxx CPU)



Easiest to do by lifting the resistor attached to pin 57 and then running a small patch wire to a 3.3V logic point (3.3V) patch, eg: FSLa pin 10, to have hardware overclock to 166Mhz FSB. A successful implementation of this done by tweakertje here. PLL Pinmod Overclocking Methods and Examples thread has more hardware overclocking info.

[angelomiguel]

Hi,

I tried the caddy from notebook elite from ebay.
I tried it on a HP NC2400 which is the same as the 2510p but one generation older with diferent chipsets.

I installed a seagate momentus 3 - 80GB 2.5" and it didn´t recognized the HD.
I tried installing XP with no sucess and the parttion manager i usually use didn´t recognize the HD also...

So for NC2400 users it´s useless!!

PLEASE, if you hve any sucess with the caddy on the 2510p please let us know something, ok?!?

At least i´m interested because i would sell my NC2400 to buy a 2510p if i could intall a 2.5" HD on it...

Originally Posted by wesselcolsen

Has anybody actually tried attaching a sata drive in the optical drive caddy yet? Does it need a pata to sata converter as well? The caddy I have bought is a sata caddy (not delivered yet) and it would seem like the way to go as I have no real use for the optical drive.

Hope it works!

[nando4]

Theory: enabling and connecting to the ICH8M native SATA port

Step 1: Enabling the native SATA controller using baredit

i. By using a 6710b bios

- flash the 6710b bios to the system. This bios enabled the sata controller AND likely provides bootup to the sata drive. For whatever reason, the 6710p bios does not boot the 2510P's master OR slave PATA drives (1.8", 2.5" in optical bay OR optical drive). The likely reason the optical drive doesn't work is because it is expecting it to be master, rather than the slave setup on the 2510P.

The way to flash the 6710b bios to the 2510P is by copying the 6710p bios.rom file as 68MSP.bin on a floppy drive, holding WIN+B on bootup to do an emergency flash recover. The bios update will prompt to do two passes to complete it. DO NOT DO TWO PASSES(!!). If you do, then likely can't reverse the process back to the original 2510P bios. Reverse by copying the 2510P bios.rom to floppy as 68DDU.bin and repeating.

The same procedure may be possible without using the floppy disk by using the a modified rompaq binary here.

ii. Manually by altering PCI configuration registers

Quick step-by-step visual instructions are shown below right. Can skip reading how/why this was done and go Step 2. Interested readers can see ICH8-M datasheet extract in italics below tells us:

The SATA1 controller is enabled/disabled via the "FD-Function Disable Register". When a function is disabled, software must not attempt to re-enable it. A disabled function can only be re-enabled by a platform reset.

LPC Interface PCI Register Address Map (LPC I/F—D31:F0)
Offset / Mnemonic / Register Name / Default / Type
F0h–F3h / RCBA /Root Complex Base Address / 00000000h / R/W

Accesses in this space must be limited to 32-(DW) bit quantities. Burst accesses are not allowed.

Using BarEDIT->Configuration Space set to Bus.0 Dev 31 Fct 0, we find the Base Address Register at 0F0 contains the string "FED90001". This tells us what the memory space is to for the FD-Function Disable Register used to enable/disable the SATA controller.

So we add 3418 to FED90001 and find the address FED93418. Bit 2 at this address has a 1 by default (disable SATA controller). Changing it to 0 then enables the SATA controller and XP will now have a new device 2828h Non-AHCI mode SATA controller appear that will automatically load drivers and appear as in the screenshot above.

To do this prior to OS boot, would require either a DOS based bootdisk with peritool to do the memory write 'pt MEM write 4 0xFED93418 0x33c0001', or using the grub2 bootloader add an entry against the OS item:

write_dword 0xFED93418 0x33c0001

Step 2: Physically connecting to a SATA port - where is the port?

Referring to the 2510P schematic, the sata0 controller would need to be physically wired to a sata device to work. Can see exactly how the ICH8M sata lines need to be configured by referring to the 6710b schematic and 6910b schematic, both systems delivered from the factory with a primary sata drive bay.

FD—Function Disable Register Offset Address: 3418–341Bh Attribute: R/W, RO Default Value: See bit description Size: 32-bit BIT 2: Serial ATA Disable 1 (SAD1) — R/W. Default is 0. 0 = The SATA controller #1 (D31:F2) is enabled. 1 = The SATA controller #1 (D31:F2) is disabled. This block is mapped into memory space, using register RCBA of the PCI-to-LPC bridge:

To do this would require connecting the PLL to the SATACLK lines on ICH8M, run 4 sata0 RX/TX lines, 3.3V and GND to your drive or 4mm runcore SATA SSD (designed for eee PCs..) pinout as per SATA pins of asus proprietory flash_con. It could sit somewhere in the chassis where there is room (eg: spare pci-e slot, pc-card slot), so the 1.8" PATA drive may be able to stay where it is That would be sensational! Or of course, could substitute the 1.8" PATA drive with a 1.8" SATA SSD like Intel X18-M or a good value ebay 1.8" Samsung SATA SSD. Right: 4 SATA I/O lines plus 3.3V and GND attached to a Samsung SSD, a working setup, as found in the macbook air forum.

[chrixx]

nando4, can you please benchmark the trnasfer rates from the hard drive in the optical bay? My 8710w only yields 15MB/s transfer rates from the same interface.

[chrixx]

Originally Posted by nando4

I am awaiting benchmarks before I consider this as a solution. Your comment is the second I've seen referring to poor performance with a SATA caddy (I presume your's is a SATA one?). For this reason my personal leanings are towards the PATA caddy with a fast 2.5" PATA drive, like the 160BG Samsung HM160JC or 250GB WD2500BEV. Those drives give very good performance and the Samsung has very low power consumption, lower than the lowest SATA drives. A workable solution if using 2.5" HDD that is limited to 250GB.

Can you identify the caddy you are using and what is the SATA-to-PATA bridge chip on it? Quite a few different one's exist Eg: Marvell, Sunplus, J-micron as seen here? The caddy would internally use something similar to this, with a different sata connector on it.

The closest thing to benchmarks I've seen is this. Though this is the second post I've seen claiming low transfer rates with a SATA caddy.. I'll see if I can find the other post on here.

From what I can make out the 8710W has a multibayII connector for the optical drive. So you have an SATA alternative to the EBAY caddy that you can buy right now (if it is 12.7mm) here. 2510P users do not. 2510P user's only option is the notebookelite (ebay) SATA product to replace their 9.5mm JAE 50 ended optical drives. That is until newmodeus release their's next month using a bridge chip "other than Marvell", as quoted by my email enquiries and will have a PATA offering as well.

So if you are using the ebay product, there are two workarounds. Enquiring/purchasing the newmodeus product, or using a PATA multibayII caddy or 2nd Hard Drive caddy for HP COMPAQ /NC6000 NX5000 and a PATA drive. The PATA caddy uses the native interface simply rewiring the interface connector, as can be seen by this.

I purchased the newmodeus standard optical bay hard drive caddy (PATA-SATA) as can be seen here http://newmodeus.com/shop/index.php?...roducts_id=179. I installed a 7k200 SATA drive in it.

The 8710w does not have a MultiBayII connector (the optical drive is fixed). It requires the JAE50 connector.

You're saying the possible bottleneck could be the PATA-SATA bridge and that a PATA-PATA version of the caddy could work better? I'm curious to know what HP's standard MultiBayII hard disk transfer rates are because the bottleneck may also be due to the own interface on the motherboard.

[chrixx]

Originally Posted by nando4

The HP MultibayII drive option I've seen on EBAY are 80GB PATA drives.

The optical bay systemboard JAE 50 connector is simply a secondary PATA IDE port. A SATA-PATA caddy needs a bridge chip to convert from one interface to the other. A PATA-PATA caddy does nothing other than rewire from 40-pin IDE to JAE50. So given there are fast PATA drives doing ~60MB/s, I would *assume* a PATA solution would likely see a better result than your 15mb/s.

I'd be double checking things like UDMA-2 is enabled against the device too. HP might be able to advise on what sort of performance their PATA caddy would give. Otherwise, perhaps ask newmodeus for advice. Better results with the SATA device would be seen if you too could identify a hidden native SATA port connector on the systemboard and wire it directly to your drive instead, as described above for the HP 2510P. ICH8-M provides 3 SATA ports.. 8710W also has an unused 24-pin socket on the systemboard as can be seen at the HP Media Services Library which *speculation mode on* could have SATA I/O pins on it. Then again, that socket could be used for diagnostic purposes as well.

There is one further variable here. What is the chipset that you are using? ICH8-M on a 2510P has a PATA controller onboard provided with it's southbridge. Google tells me 8710W uses PM965, so it should also be using ICH8-M. I know that Intel was planning to ditch the PATA interface and if so, then the native ICH controller is SATA and HP would have had to incorporate a SATA-PATA bridge on the systemboard to run legacy PATA devices like optical drives.

Oh.. can you also identify the SATA-to-PATA bridge chip on your caddy? It will be the biggest chip on there.. Marvell, Sunplus, Acard, Jmicron. newmodeus were tight-lipped about what they are using in their up coming 9.5mm caddy.

The 8710w uses the ICH8-M chipset.
The SATA-to-PATA bridge chip on the caddy doesn't seem to have a brand (it's all blacked out).

[nando4]

INFO: Performance of SATA SSD in a 2nd drive caddy [Lenovo Ultrabay PATA]

Below is CrystalDIskMark benchmarks comparing performance of a SATA Mtron SLC 3500 100/100 SSD using a SATA-to-PATA bridge adapter versus a native SATA port on a Thinkpad T60 using a ICH7M ATA100/UDMA5 optical drive interface. Original posting is here. CrystalDiskMark: SATA MTRON 3500 SLC 100/100 ..........Read..MB/s...................Write..MB/s Seq.....84.28:94.98 (88.7%)......55.92:86.95 (64.3%) 512k....77.42:94.68 (81.8%)......21.22:22.49 (94.4%) 4k.......18.73:19.78 (94.7%)........1.28:2.81...(45.6%) Left of colon: SATA-to-PATA ultrabay optical bay caddy Right of colon: native SATA interface The bridge chip does degrade read performance by 12%. That would be OK by me. I am curiously awaiting benchmarks of the equivalent HP 2510P SATA 2nd drive caddy and SATA HDD/SSD. Anyone??

[oward]

All,

I've received today the notebookelite sata caddy and can confirm it's not working in the 2510p ...
I've made the test with two different hard drives with same results (Toshiba and Samsung).
The second harddrive is neither detected by BIOS or Windows
Sometimes for the Toshiba, I've got a fixed red LED on the caddy waiting long seconds for BIOS to boot, ending with a "Disk 0 startup error" anyway.
Seems either the SATA bridge has some issues or that BIOS is locked to detect only optical drive ?
Anyway, I've already switched my internal 1.8" harddisk to a 5400 rpm Samsung HS122JC and fired up a 4GB RAM module into my 2510p
This combination gives me quite satisfactory results under Vista