port expanders would not help and probably using multiple props would cause more synchronization so I don't think you can fetch a long in a single clock cycle (at least more than one). Probably need a clock rate of 2ns to sample memory at 10 ns - that would allow 5 instructions and that would be very tight.
For bust reads/writes (frame buffer access) it is not quite so simple - using multiple cogs, On Morpheus, I am currently doing 20MB/sec with 55ns ram using 3 cogs, and if I use faster memory (and two more cogs) I can easily do 40MB/sec - assuming I use 20ns or faster SOJ36 fast memory. As soon as I can take a break from current consulting (and PropellerBasic) I intend to make some fast modules for myself and prove that 40MB/sec and 50MB/sec work.
I also believe I can do 40MB/sec with mctrivia's module (and the Morpheus adapter for it).
Theoretically I've figured out how to do 80MB/sec and 100MB/sec, but I would probably need 8.4ns ram for that (and five cogs).
Yoda said...
Jazzed said...
You have no proof of that Ray
I think simple math proves that with a 10ns clock you need one clock pulse to "clock the data" which is going to be 2 instructions to go low then high and back down. Next you have to set up a new address which will take several instructions and repeat the clocking. If you want 4 bytes at a time you would have to read all the data from 4 byte wide srams = 32 bits - you are out of pins on a propeller. Seems like the laws of physics say this is not really possible especially for long accesses or am I missing something?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ www.mikronauts.com Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full Morpheusdual Prop SBC w/ 512KB kit $119.95, Mem+2MB memory IO board kit $89.95, both kits $189.95 Propteus and Proteus for Propeller prototyping 6.250MHz custom Crystals run Propellers at 100MHz Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
I think simple math proves that with a 10ns clock you need one clock pulse to "clock the data" which is going to be 2 instructions to go low then high and back down. Next you have to set up a new address which will take several instructions and repeat the clocking. If you want 4 bytes at a time you would have to read all the data from 4 byte wide srams = 32 bits - you are out of pins on a propeller. Seems like the laws of physics say this is not really possible especially for long accesses or am I missing something?
Propeller offers parallel execution and self-modifying code. It takes a while to understand the benefits.
5 COG + Counters Algorithm:
1. start 4 data collector cogs that point one each to a byte of the long to read into hub memory
2. start an address and long hub read cog
3. set address to bus ... collector cogs wait for non-zero on address bus before starting byte read
- since byte reads are accomplished according to the cog enumeration, the read is back to back
- address cog increments lower address bits with counter a/b outputs for collector cogs
4. address and long read cog pulls long out of hub to lmm cog
Long read rate 1 long/50ns or 20MB/s at 80MHz
LMM data rate 2.5MIPS assuming long at a time fetch/execute.
I'm working on a test of this now ... instead of other important things [noparse]:)[/noparse]
Will update later as time permits.
add addr,#cog_offset
top:
rdlong addr,xlmm_pc wz
if_z jmp #top
mov temp,ina
wrbyte temp,addr ' too bad wrbyte ina,addr does not work!
jmp #top
Requires 32 cycles per long (ideal full stride) = 80M/32 = 2.5 XLMM mips
Needs four reader cogs, one address cog, and one XLMM cog.
Neat.
jazzed said...
Yoda said...
Jazzed said...
You have no proof of that Ray
I think simple math proves that with a 10ns clock you need one clock pulse to "clock the data" which is going to be 2 instructions to go low then high and back down. Next you have to set up a new address which will take several instructions and repeat the clocking. If you want 4 bytes at a time you would have to read all the data from 4 byte wide srams = 32 bits - you are out of pins on a propeller. Seems like the laws of physics say this is not really possible especially for long accesses or am I missing something?
Propeller offers parallel execution and self-modifying code. It takes a while to understand the benefits.
5 COG + Counters Algorithm:
1. start 4 data collector cogs that point one each to a byte of the long to read into hub memory
2. start an address and long hub read cog
3. set address to bus ... collector cogs wait for non-zero on address bus before starting byte read
- since byte reads are accomplished according to the cog enumeration, the read is back to back
- address cog increments lower address bits with counter a/b outputs for collector cogs
4. address and long read cog pulls long out of hub to lmm cog
Long read rate 1 long/50ns or 20MB/s at 80MHz
LMM data rate 2.5MIPS assuming long at a time fetch/execute.
I'm working on a test of this now ... instead of other important things [noparse]:)[/noparse]
Will update later as time permits.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ www.mikronauts.com Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full Morpheusdual Prop SBC w/ 512KB kit $119.95, Mem+2MB memory IO board kit $89.95, both kits $189.95 Propteus and Proteus for Propeller prototyping 6.250MHz custom Crystals run Propellers at 100MHz Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
Jazzed - think you left out a few steps (instructions). I think it is more like what Bill suggests which does not get you to 10ns per read as you said could be done. Even if you accept what you wrote you don't get 10ns per long. Granted there are fast and clever ways to do reads - I just don't think 10ns per long read is feasible. In you scheme you need more than 32 I/O pins I believe which means multiple propellers and synchronization?
Yoda said...
Jazzed - think you left out a few steps (instructions). I think it is more like what Bill suggests which does not get you to 10ns per read as you said could be done. Even if you accept what you wrote you don't get 10ns per long. Granted there are fast and clever ways to do reads - I just don't think 10ns per long read is feasible. In you scheme you need more than 32 I/O pins I believe which means multiple propellers and synchronization?
I'm afraid you have mis-understood. No one but you have suggested reading 10ns per long.
Ok - I guess I am confused - I interpret 10ns access as wanting to fetch in 10ns and you were talking longs - so I guess I misunderstood what you are trying to do then
Sure. Getting one long per LMM 2.5MIPS cycle is the goal. To do that, one needs to read a byte every 10ns or so.
The algorithm I posted is difficult to keep the lower 2 address lines in sync and I've added another cog for now.
Don't want to OT this thread too much, so I'll keep quiet now until I have a solid demo or say uncle.
I have to take a break the rest of the day for other commitments though [noparse]:)[/noparse]
The pcbs have arrived but I am interstate. I am waiting for a photo of them and I will upload as soon as I can.
The parts have all arrived in TX and hopefully will be shipped to Oz today. Tubular is organising this (many thanks) and they will also arrive interstate next week and then onto me, whereever I am.
I hope to have the bootloader (eeprom code) ready and tested before I have the parts. Likewise, I hope to have the drivers done by then.
As I disclosed before, the address lines A0-18 are on prop pins P0-18 and the data lines D0-7 are on P24-31.
For the couple of people waiting on TrBlades with microSD and FT232RL, these parts are in the RamBlade shipment.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
Cluso99 said...
I have the pcbs in my hands. Just waiting on the parts now
Here are some photos - notice how small it is
Some mounting options shown for the Prop ProtoBoard (my proto has other things loaded).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ www.mikronauts.com Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full Morpheusdual Prop SBC w/ 512KB kit $119.95, Mem+2MB memory IO board kit $89.95, both kits $189.95 Propteus and Proteus for Propeller prototyping 6.250MHz custom Crystals run Propellers at 100MHz Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Nothing is impossible, there are only different degrees of difficulty. For every stupid question there is at least one intelligent answer. Don't guess - ask instead. If you don't ask you won't know. If your gonna construct something, make it·as simple as·possible yet as versatile as posible.
I only really discovered just how small it was a week ago when I printed it out on paper to use for sizing another design. It is amazing when you lay a tiny pcb out that you do not really grasp the physical size when looking at on a big screen.
I mentioned I also want the RamBlade to be able to run self-contained. Just a PS2 keyboard, TV (video), and power pack (5V regulated to power the keyboard). Well, I have some news...
I have 1-pin video running - requires 1 resistor to be added to the pcb and a video cable. On my LCD I can get 80x24 resolution in monochrome (96MHz clock). There is a little glimmering, but it is acceptable. Thanks to ericball for the 1-pin driver and Phil's and OBC's inspiration to get it written. There is a bug to be fixed, and terminal code to locate. http://forums.parallax.com/forums/default.aspx?f=25&m=366805
I can do 2-pin PS2 keyboard with 4 resistors and stealing a SRAM pin. However, looking at the keyboard code in the OBEX, I modified it and Ariba has tested it and found that we do not have to send data to the keyboard. Obviously, the LEDs won't work. http://forums.parallax.com/forums/default.aspx?f=25&m=407823
So, now I know that we can just receive, I will try to implement a 1-pin keyboard interface based on the recent thread using an RC network. Obviously it is running the other way, so I don't have control of the keyboard. We will see??·http://forums.parallax.com/showthread.php?p=860512
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
Jazzed: Thanks. I am just redoing the driver in the light of recent timing info. IIRC reads were 4 instructions and writes 7. (always the other pc !!!) I may be able to do reads in 3???
I had trouble finding the thread, but I did end up locating it (code on paper interstate). I think I searched for "external sram XMM" ??
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
Armed with the new knowledge, srams with access times <25ns (for 100MHz) may be read faster than 55ns using a single cog. This depends on the code, because it may be possible to rearrange the instructions without penalty. I can increment the address where an access delay is required without penalty.
There will be no difference in writing though using a single cog.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
Cluso99 said...
Jazzed: Here is the link with the code. ...
I couldn't find your link. Thought I would offer up a simple readlong / lmm execute loop.
Maybe you can offer this for XLMM'ers if you haven't already derived similar code. Is there an optimization?
'--------------------------------------------------------------------------------
' d24:31 a0:23
' DIRA for read is $00ff_fffc
' OUTA holds the PC
' Effects of OUTA bits with devices are unknown
' Counter setup for address stepping tbd but should work with even sampling points.
'
nexti mov phsa, _phsa ' counters start with high so we
mov phsb, _phsb ' get 1100 and 1010 or 3210
nop ' Use fast SRAM to remove this.
mov inst, ina ' AAAAAAAA_axxxxxxx_xxxxxxxx_xxxxxxxx
ror inst, #9 ' xxxxxxxx_xxxxxxxx_xxxxxxxA_AAAAAAAx
mov dat1, ina ' BBBBBBBB_bxxxxxxx_xxxxxxxx_xxxxxxxx
movs inst, dat1 ' xxxxxxxx_xxxxxxxA_AAAAAAAB_BBBBBBBb
mov dat2, ina ' CCCCCCCC_cxxxxxxx_xxxxxxxx_xxxxxxxx
ror inst, #8 ' xxxxxxxA_AAAAAAAB_BBBBBBBb_xxxxxxxx
mov dat3, ina ' DDDDDDDD_dxxxxxxx_xxxxxxxx_xxxxxxxx
movs inst, dat2 ' xxxxxxxA_AAAAAAAB_BBBBBBBC_CCCCCCCx
ror inst, #7 ' AAAAAAAA_BBBBBBBB_CCCCCCCC_xxxxxxxx
shl insta, #1 wc ' AAAAAAAB_BBBBBBBC_CCCCCCCx_xxxxxxxx C=A7
movs inst, dat3 ' AAAAAAAB_BBBBBBBC_CCCCCCCD_DDDDDDDD
rcr inst, #1 ' AAAAAAAA_BBBBBBBB_CCCCCCCC_DDDDDDDD
add outa, #4
inst nop
jmp #nexti ' 18 COG / XMM instruction = 1.38 MIPS (100MHz)
The dinky ram chip I mentioned is TSOP, about 40% the area of the one on the board (pessumably yours is still a '4008) . I doubt I shall ever use it as it is just so ********* small
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
TSOP & sTSOP are 0.5mm pin spacing whereas the SRAM I use is the SOP which is 1.27mm pin spacing (like the rectangular chip beside the qfp in your pic).
The SRAM was the easiest part to solder, and the microSD. The hardest are the 4x10K resistor packs, the SOT23-6 and the SOIC (eeprom). As you can see from the photos, there was no point in using the much harder to solder TSOP/sTSOP SRAMs. The pcb could be smaller if not for the extra width and length required by the multipurpose connector.
I am running tests on the RamBlade now
Tubular: No, the photo is out of focus to hide my bad nightime soldering·· I just couldn't wait until the morning!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
We have an old hot air station at work. When I get a chance I find an old pcb and lift off a chip or two, and then practice sticking them back down again. When just practicing they usually go back on beautifully, if it' s to be used properly, or somebody else will look at it, then they are a right pig. Gelled flux seems to be the best but a few years ago one of the guys got hold of some liquid flux, from Philips I think, it was fantastic but doesn't meet the new laws. Lead free solder is a joke too.
Ps the qfp and the other chip are a Prop and a '4008, for future playings.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
Post Edited (Toby Seckshund) : 12/12/2009 2:39:42 PM GMT
Comments
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
JMH
For bust reads/writes (frame buffer access) it is not quite so simple - using multiple cogs, On Morpheus, I am currently doing 20MB/sec with 55ns ram using 3 cogs, and if I use faster memory (and two more cogs) I can easily do 40MB/sec - assuming I use 20ns or faster SOJ36 fast memory. As soon as I can take a break from current consulting (and PropellerBasic) I intend to make some fast modules for myself and prove that 40MB/sec and 50MB/sec work.
I also believe I can do 40MB/sec with mctrivia's module (and the Morpheus adapter for it).
Theoretically I've figured out how to do 80MB/sec and 100MB/sec, but I would probably need 8.4ns ram for that (and five cogs).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.mikronauts.com Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full
Morpheusdual Prop SBC w/ 512KB kit $119.95, Mem+2MB memory IO board kit $89.95, both kits $189.95
Propteus and Proteus for Propeller prototyping 6.250MHz custom Crystals run Propellers at 100MHz
Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
5 COG + Counters Algorithm:
1. start 4 data collector cogs that point one each to a byte of the long to read into hub memory
2. start an address and long hub read cog
3. set address to bus ... collector cogs wait for non-zero on address bus before starting byte read
- since byte reads are accomplished according to the cog enumeration, the read is back to back
- address cog increments lower address bits with counter a/b outputs for collector cogs
4. address and long read cog pulls long out of hub to lmm cog
Long read rate 1 long/50ns or 20MB/s at 80MHz
LMM data rate 2.5MIPS assuming long at a time fetch/execute.
I'm working on a test of this now ... instead of other important things [noparse]:)[/noparse]
Will update later as time permits.
Something like:
add addr,#cog_offset top: rdlong addr,xlmm_pc wz if_z jmp #top mov temp,ina wrbyte temp,addr ' too bad wrbyte ina,addr does not work! jmp #topRequires 32 cycles per long (ideal full stride) = 80M/32 = 2.5 XLMM mips
Needs four reader cogs, one address cog, and one XLMM cog.
Neat.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.mikronauts.com Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full
Morpheusdual Prop SBC w/ 512KB kit $119.95, Mem+2MB memory IO board kit $89.95, both kits $189.95
Propteus and Proteus for Propeller prototyping 6.250MHz custom Crystals run Propellers at 100MHz
Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
I'm afraid you have mis-understood. No one but you have suggested reading 10ns per long.
The statement was: "10nS access is not possible!"
The algorithm I posted is difficult to keep the lower 2 address lines in sync and I've added another cog for now.
Don't want to OT this thread too much, so I'll keep quiet now until I have a solid demo or say uncle.
I have to take a break the rest of the day for other commitments though [noparse]:)[/noparse]
Cheers.
I'll do it now...
BTW the RamBlade pcbs shipped were shipped to me yesterday
There are 7 orders, all with microSD cards. This was surprising to me.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
The pcbs have arrived but I am interstate. I am waiting for a photo of them and I will upload as soon as I can.
The parts have all arrived in TX and hopefully will be shipped to Oz today. Tubular is organising this (many thanks) and they will also arrive interstate next week and then onto me, whereever I am.
I hope to have the bootloader (eeprom code) ready and tested before I have the parts. Likewise, I hope to have the drivers done by then.
As I disclosed before, the address lines A0-18 are on prop pins P0-18 and the data lines D0-7 are on P24-31.
For the couple of people waiting on TrBlades with microSD and FT232RL, these parts are in the RamBlade shipment.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Here are some photos - notice how small it is
Some mounting options shown for the Prop ProtoBoard·(my proto has other things loaded).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Looks really good...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.mikronauts.com Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full
Morpheusdual Prop SBC w/ 512KB kit $119.95, Mem+2MB memory IO board kit $89.95, both kits $189.95
Propteus and Proteus for Propeller prototyping 6.250MHz custom Crystals run Propellers at 100MHz
Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.smarthome.viviti.com/propeller
VERY nice board.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Nothing is impossible, there are only different degrees of difficulty.
For every stupid question there is at least one intelligent answer.
Don't guess - ask instead.
If you don't ask you won't know.
If your gonna construct something, make it·as simple as·possible yet as versatile as posible.
Sapieha
Coley
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
PropGFX - The home of the Hybrid Development System and PropGFX Lite
I only really discovered just how small it was a week ago when I printed it out on paper to use for sizing another design. It is amazing when you lay a tiny pcb out that you do not really grasp the physical size when looking at on a big screen.
I mentioned I also want the RamBlade to be able to run self-contained. Just a PS2 keyboard, TV (video), and power pack (5V regulated to power the keyboard). Well, I have some news...
I have 1-pin video running - requires 1 resistor to be added to the pcb and a video cable. On my LCD I can get 80x24 resolution in monochrome (96MHz clock). There is a little glimmering, but it is acceptable. Thanks to ericball for the 1-pin driver and Phil's and OBC's inspiration to get it written. There is a bug to be fixed, and terminal code to locate. http://forums.parallax.com/forums/default.aspx?f=25&m=366805
I can do 2-pin PS2 keyboard with 4 resistors and stealing a SRAM pin. However, looking at the keyboard code in the OBEX, I modified it and Ariba has tested it and found that we do not have to send data to the keyboard. Obviously, the LEDs won't work. http://forums.parallax.com/forums/default.aspx?f=25&m=407823
So, now I know that we can just receive, I will try to implement a 1-pin keyboard interface based on the recent thread using an RC network. Obviously it is running the other way, so I don't have control of the keyboard. We will see??· http://forums.parallax.com/showthread.php?p=860512
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Post Edited (Cluso99) : 12/8/2009 10:37:10 AM GMT
The 1-pin keyboard will almost certainly work
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Congrats.
I had trouble finding the thread, but I did end up locating it (code on paper interstate). I think I searched for "external sram XMM" ??
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Bursting data to/from Cog RAM- 6.6MB/sec using 9 pins?·· http://forums.parallax.com/showthread.php?p=805698
Armed with the new knowledge, srams with access times <25ns (for 100MHz) may be read faster than 55ns using a single cog. This depends on the code, because it may be possible to rearrange the instructions without penalty. I can increment the address where an access delay is required without penalty.
There will be no difference in writing though using a single cog.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Post Edited (Cluso99) : 12/14/2009 4:56:31 AM GMT
Please don't look too closely at the pics·above. Pics below are better
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Post Edited (Cluso99) : 1/2/2010 10:06:30 AM GMT
Maybe you can offer this for XLMM'ers if you haven't already derived similar code. Is there an optimization?
'-------------------------------------------------------------------------------- ' d24:31 a0:23 ' DIRA for read is $00ff_fffc ' OUTA holds the PC ' Effects of OUTA bits with devices are unknown ' Counter setup for address stepping tbd but should work with even sampling points. ' nexti mov phsa, _phsa ' counters start with high so we mov phsb, _phsb ' get 1100 and 1010 or 3210 nop ' Use fast SRAM to remove this. mov inst, ina ' AAAAAAAA_axxxxxxx_xxxxxxxx_xxxxxxxx ror inst, #9 ' xxxxxxxx_xxxxxxxx_xxxxxxxA_AAAAAAAx mov dat1, ina ' BBBBBBBB_bxxxxxxx_xxxxxxxx_xxxxxxxx movs inst, dat1 ' xxxxxxxx_xxxxxxxA_AAAAAAAB_BBBBBBBb mov dat2, ina ' CCCCCCCC_cxxxxxxx_xxxxxxxx_xxxxxxxx ror inst, #8 ' xxxxxxxA_AAAAAAAB_BBBBBBBb_xxxxxxxx mov dat3, ina ' DDDDDDDD_dxxxxxxx_xxxxxxxx_xxxxxxxx movs inst, dat2 ' xxxxxxxA_AAAAAAAB_BBBBBBBC_CCCCCCCx ror inst, #7 ' AAAAAAAA_BBBBBBBB_CCCCCCCC_xxxxxxxx shl insta, #1 wc ' AAAAAAAB_BBBBBBBC_CCCCCCCx_xxxxxxxx C=A7 movs inst, dat3 ' AAAAAAAB_BBBBBBBC_CCCCCCCD_DDDDDDDD rcr inst, #1 ' AAAAAAAA_BBBBBBBB_CCCCCCCC_DDDDDDDD add outa, #4 inst nop jmp #nexti ' 18 COG / XMM instruction = 1.38 MIPS (100MHz)Post Edited (jazzed) : 12/11/2009 6:36:35 PM GMT
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.smarthome.viviti.com/propeller
Nothing done last night - a little tipsy [noparse]:)[/noparse]
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
I assumed that, from the blurred photo...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
The SRAM was the easiest part to solder, and the microSD. The hardest are the 4x10K resistor packs, the SOT23-6 and the SOIC (eeprom). As you can see from the photos, there was no point in using the much harder to solder TSOP/sTSOP SRAMs. The pcb could be smaller if not for the extra width and length required by the multipurpose connector.
I am running tests on the RamBlade now
Tubular: No, the photo is out of focus to hide my bad nightime soldering·
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)
· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
Post Edited (Cluso99) : 12/12/2009 9:39:45 AM GMT
Ps the qfp and the other chip are a Prop and a '4008, for future playings.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
Post Edited (Toby Seckshund) : 12/12/2009 2:39:42 PM GMT