Accelerated Simulspace and Speed

16 posts / 0 new
Last post
Lorsa Lorsa's picture
Accelerated Simulspace and Speed

I apologise for my lack of eloquence, I am writing this before having eaten and drinking coffee. In any case, I should be able to get my question across.

One thing that I have quite figured out is how it is possible to have simulspace acceleration to x60 but mental speed is still limited to 4. Yes, you can get a few extra actions through multi-tasking or the like, but that is pretty much equivalent to forking. The mental speed is still limited to 4.

How does that make sense? If you can accelerate a computer so much that your subjective time os 60 times faster, shouldn't you be able to, in a cyberbrain or as an infomorph, to get a mental speed of 60 or so? How come a simulspace is any different?

Also, since you can be sleeved into a biomorph and access a simulspace, there's no reason why the mental speed of those should be limited as well.

Anyone that has an answer to this problem?

Lorsa is a Forum moderator

Red text is for moderator stuff

LatwPIAT LatwPIAT's picture
My general answer to this

My general answer to this question is that the rules are nonsensical for the reasons you have provided, and should be replaced with more sensible ones:

Step 1: Extra mental actions don't represent speed, they represent parallelism.

Step 2: With a biological brain, physical constraints limit the ability of simulspace acceleration to some arbitrary very low limit, like x1.5 or x2. With a dedicated simulspace station that basically temporarily installs Mental Speed in the morph's brain through something akin to an Ego Bridge, this can be increased to x10 (taking the reading-speed increase from Mental Speed as a guide). If you actually have Mental Speed, you can remotely connect to x10 simulspaces. Cyberbrains and Ectos have similar limitations to those of biological brain, but with specialized hardware (Mental Speed), they can connect to and use 10x simulspaces. To get greater than x10 simulspace acceleration, you need to sleeve into an infomorph and run on a server with a very high clock speed, because (cyber)brains can't interface faster than x10.

@-rep +2
C-rep +1

Armoured Armoured's picture
Short answer? Game balance,

Short answer? Game balance, I would guess. The x60 speed multiplier on simulspaces is great as a narrative tool, with what would take months of down-time in a normal game taking a few hours in THE AWESOME FUTURE! Eclipse Phase setting. Plus its useful for (effectively) instant meetings, interrogations, psychosurgery, and the like, keeping the story flow fast.

Its kept effectively separate from realspace interactions, I think, to keep speed actually relevant. If cyberbrains could push up to x60, or even x10, everyone else would be irritants, not real opponents. Its a compromise to let us 21st century humans understand the combat. Speed is already a point of contention with many people, often being the real primary decider of combat.

If you want in-universe excuses as to why cyberbrains can't run at x60, the server hardware might still have to be too big for a morph to carry. People in biomorphs may be able to use x60 simulspace, but maybe the "connection" process is actually a mini ego compression/upload, with your memories updated when you close the connection. The connection process could take a while to complete, making it unusable in combat.

Look out for someone hitting you with a EMP grenade when you are connected! It might not be a Matrix-style unplugging insta-kill, but its still going to suck.

ThatWhichNeverWas ThatWhichNeverWas's picture
If you are Exhuman, disregrard the following:

The simple answer is hardware. The existence of the x60 limitation on acceleration means that you can't just accelerate a mind however you want; you need specialized systems to do it otherwise you get... problems.
I'd say that the x60 limit is actually pretty advanced, and most people if they accelerate at all max out at x30.
The mental speed implant, along with speed enhancements, means you can get an effective x12, so beyond that I'd say you really need a dedicated server modified to support it.
I'd also say that you can't use a server for the speed and jam a bot because the discrepancy starts causing the aforementioned problems.
Alternatively, x12 is a limit on "perception"... you can go faster, but it manifests as increased intelligence rather than time dilation.

In the past we've had to compensate for weaknesses, finding quick solutions that only benefit a few.
But what if we never need to feel weak or morally conflicted again?

ORCACommander ORCACommander's picture
maybe a bit of the opposite

maybe a bit of the opposite since they are slowing down time

https://wiki.eveonline.com/en/wiki/Time_Dilation

http://community.eveonline.com/news/dev-blogs/the-bloodbath-of-b-r5rb/

but ya there are hardware problems and the biggest thing i think is that Simspaces are self contained. they do not interact with the real world which unshackles them. think of how an ego being both in a 1x environment and a 60x environment would stress out and shred at the contradictory sensory inputs.. Another good example is DF worldgen. sure we could make the history in real time but hat would be rather tedious. unstead it just runs he sim until the paramaters are met or until and abort command is recieved.

Lorsa Lorsa's picture
Thanks for all the replies

Thanks for all the replies everyone!

LatwPIAT wrote:
My general answer to this question is that the rules are nonsensical for the reasons you have provided, and should be replaced with more sensible ones:

Step 1: Extra mental actions don't represent speed, they represent parallelism.

Step 2: With a biological brain, physical constraints limit the ability of simulspace acceleration to some arbitrary very low limit, like x1.5 or x2. With a dedicated simulspace station that basically temporarily installs Mental Speed in the morph's brain through something akin to an Ego Bridge, this can be increased to x10 (taking the reading-speed increase from Mental Speed as a guide). If you actually have Mental Speed, you can remotely connect to x10 simulspaces. Cyberbrains and Ectos have similar limitations to those of biological brain, but with specialized hardware (Mental Speed), they can connect to and use 10x simulspaces. To get greater than x10 simulspace acceleration, you need to sleeve into an infomorph and run on a server with a very high clock speed, because (cyber)brains can't interface faster than x10.

Especially thanks for this LatwPIAT. Very good suggestions and I'll definitely use them!

Lorsa is a Forum moderator

Red text is for moderator stuff

Pyrite Pyrite's picture
I usually assume that you can

I usually assume that you can't get above 4X without uploading yourself directly into the server, though 10X with mental speed wouldn't be a bad compromise.

'No language is justly studied merely as an aid to other purposes. It will in fact better serve other purposes, philological or historical, when it is studied for love, for itself.' --J.R.R. Tolkien

Chernoborg Chernoborg's picture
I generally agree with

I generally agree with LatwPIAT, I'd just add a few things.

If the background of the Lost is any indication, a biological brain could run at 6X speeds for months. Assuming that the simulspace feeds were at the maximums of a brains ability to assimilate the information. Since the Futura doesn't come with neurachem or any other speed boosts that could be our baseline. Although the healing vat they were grown in could have enabled sustained higher speeds than ordinary- could be something for the hackers equipment list!

A cyberbrain has to run the brain simulation and the hardwired hacking protections that are standard which may account for the lower than expected speeds. Also, nanotech/bioware altered nervous systems allow for about 10X faster perceptions ( see Mental Speed) so that would be a good augmented limit.

For infomorphs there could be several options for running at the 60X maximum. Specialized servers like at Glitch , low traffic environments to allow more processing resources to be allocated, and running a virtual cyberbrain with the hacking protections off are some of the ways that come to mind. Probably all of the above have to be in place for full capability.

Current Status: Highly Distracted building Gatecrashing systems in Universe Sandbox!

ORCACommander ORCACommander's picture
i was under the impression

i was under the impression that the futura project separated the ego from the futura morph to use in their rapid sim space

Chernoborg Chernoborg's picture
I remember some debate on

I remember some debate on this a while ago. My interpretation of

Quote:
After the sample was selected, all subjects were sleeved into our fast-growth futura brand biomorph bodies and inducted into customized simulspace accelerated learning environments.
is that they were sleeved into approximately same aged morphs and then fast grown with the simulspace providing sensory input as fast as their developing brains could absorb.

Current Status: Highly Distracted building Gatecrashing systems in Universe Sandbox!

branford branford's picture
Chernoborg wrote:I remember

Chernoborg wrote:
I remember some debate on this a while ago. My interpretation of
Quote:
After the sample was selected, all subjects were sleeved into our fast-growth futura brand biomorph bodies and inducted into customized simulspace accelerated learning environments.
is that they were sleeved into approximately same aged morphs and then fast grown with the simulspace providing sensory input as fast as their developing brains could absorb.

Who could have foreseen any trouble with that idea . . .? /sarc.

Lorsa Lorsa's picture
branford wrote:Chernoborg

branford wrote:
Chernoborg wrote:
I remember some debate on this a while ago. My interpretation of
Quote:
After the sample was selected, all subjects were sleeved into our fast-growth futura brand biomorph bodies and inducted into customized simulspace accelerated learning environments.
is that they were sleeved into approximately same aged morphs and then fast grown with the simulspace providing sensory input as fast as their developing brains could absorb.

Who could have foreseen any trouble with that idea . . .? /sarc.

The research guy who got fired for pointing out obvious flaws with the plan I reckon.

Lorsa is a Forum moderator

Red text is for moderator stuff

LatwPIAT LatwPIAT's picture
Chernoborg wrote:I generally

Chernoborg wrote:
I generally agree with LatwPIAT, I'd just add a few things.

If the background of the Lost is any indication, a biological brain could run at 6X speeds for months. Assuming that the simulspace feeds were at the maximums of a brains ability to assimilate the information. Since the Futura doesn't come with neurachem or any other speed boosts that could be our baseline. Although the healing vat they were grown in could have enabled sustained higher speeds than ordinary- could be something for the hackers equipment list!

I think I'd go with the healing vats being special here; it'd basically be like the Ego Bridge/Mental Speed thing I suggested.

Chernoborg wrote:
For infomorphs there could be several options for running at the 60X maximum. Specialized servers like at Glitch , low traffic environments to allow more processing resources to be allocated, and running a virtual cyberbrain with the hacking protections off are some of the ways that come to mind. Probably all of the above have to be in place for full capability.

60x[Ego simulation demands] could simply be the hardware limitations of AF10 computers. Servers capable of running multiple Egos can use parallelism to run several at once, but - for example - the RAM-fetching speeds mean that a parallel solution can't be used to run a whole-brain emulation at faster than 60x, or supercomputer processing speeds are currently peaking at 60x[Ego simulation demands]. The latter would imply that - other hardware concerns being insignificant - the maximum simulspace acceleration would grow to 120x by AF11.5 assuming Moore's Law holds.

@-rep +2
C-rep +1

Lorsa Lorsa's picture
LatwPIAT wrote:60x[Ego

LatwPIAT wrote:
60x[Ego simulation demands] could simply be the hardware limitations of AF10 computers. Servers capable of running multiple Egos can use parallelism to run several at once, but - for example - the RAM-fetching speeds mean that a parallel solution can't be used to run a whole-brain emulation at faster than 60x, or supercomputer processing speeds are currently peaking at 60x[Ego simulation demands]. The latter would imply that - other hardware concerns being insignificant - the maximum simulspace acceleration would grow to 120x by AF11.5 assuming Moore's Law holds.

I'm certainly not going to interfere with anyone who wants to go by that, but from what I know Moore's Law is already breaking down, or on the verge of doing so. Once we move into other transistor technology than MOSFETs made with Optical Lithography I'm guessing we'll have another "law" entirely.

Lorsa is a Forum moderator

Red text is for moderator stuff

LatwPIAT LatwPIAT's picture
Lorsa wrote:I'm certainly not

Lorsa wrote:
I'm certainly not going to interfere with anyone who wants to go by that, but from what I know Moore's Law is already breaking down, or on the verge of doing so. Once we move into other transistor technology than MOSFETs made with Optical Lithography I'm guessing we'll have another "law" entirely.

The naive projection of continued exponential growth of computational power until we hit the Berkenstein limit is something I tend to laugh at and ridicule, but given that there's very little information available on the nature of computing in EP except rough guidelines, I tend to make the naive assumption that Moore's Law holds, and preface all statements dependent upon that assumption with "assuming Moore's Law holds".

More interesting is probably the absolute limit set by the Landauer limit; making some very gross assumptions about computing efficiency (FLOPs per cycle are at maximum efficiency, no reversible computing, no overhead costs), if you were to run a computer using the output of the four nuclear reactors in a Nimitz-class aircraft carrier - and obscene amount of power - it could not possibly run faster than 10^30 FLOPS. Using the power-output of an average US household, it's impossible to get a computer to run faster than 10^23 FLOPS.

Actually, given the described capabilities of EP hardware, and knowledge of the level of power consumption or power-production available, it would be possible to calculate the computational power available.

Which I can do. It's dead simple. The core book says that batteries have 25 times the capacity of modern-day batteries, and last for 100-500 hours. Modern-day Lithium-ion batteries have a capacity of about 20kW/kg, and a density of about 2.736 g/cm^3. With a known volume, density, and power-to-weight ratio, it is then possible to calculate the power output of an EP battery. An Ecto is the size of a credit card, and credit cards have sizes rigidly set by an ISO standard. Making the silly assumption that an Ecto is 100% battery, the energy consumption of an Ecto cannot be more than about 5kW. Using the Landauer limit at room temperature, this gives an Ecto a maximum computational power of 10^23 FLOPS. It is likely to be much, much less than that, because the actual processor has to fit somewhere in the Ecto too.

What this does tell us is that a Whole Brain Emulation in EP cannot require more than 10^23 FLOPS to run, which according to Areamontanous paper indicates that nothing more advanced than an electrophysiological Whole Brain Emulation is in use.

Mind, I'm assuming that the devs did not do this level of nitpicking calculations (or use my assumptions), and therefore may say or have assumed that a more advanced WBE model is in use, or something like that.

/derail

@-rep +2
C-rep +1

Lorsa Lorsa's picture
Thanks for the detailed

Thanks for the detailed insight LatwPIAT. As you say, it would be very surprising if computational power in EP was operating at the Landauer limit. Or maybe I should say if transhuman computers worked that way.

I am not aware of the requirements for Whole Brain Emulations, although I am assuming that current-day theories would list it as being much lower than the 10^23 FLOPS that yuo mentioned.

Personally I am not going to assume Moore's Law in EP, mostly because I don't want that kind of exponential growth in my campaigns, but also because it is much more of a description than a law. There is nothing intrinsically physical about it, and applying it to other techniques of computer manufacturing than optical lithography is a stretch at best. I've always seen Moore's Law as being descriptive, not predictive.

Lorsa is a Forum moderator

Red text is for moderator stuff