[sims] Sims Discover

[AyatollahPies]

I suppose being on EduGeek you're interested in the technology. I'm interested in what something can do and I've seen enough revelations in schools to know that this is a very powerful tool.

 

The development team's brief was not to make it run on a hairdryer; it was to produce a very powerful tool that was really easy to use. You do need to understand assessment result sets to get valuable information out however or you will be reduced to finding out how many Italians eat sandwiches!

 

As it happens in will run on fairly ordinary kit but integrated graphics are of variable quality.

 

Ah, but did you have everyone's favourite resource hogging Anti-Virus running in the background?

 

[LosOjos]

I suppose being on EduGeek you're interested in the technology. I'm interested in what something can do and I've seen enough revelations in schools to know that this is a very powerful tool.

 

The development team's brief was not to make it run on a hairdryer; it was to produce a very powerful tool that was really easy to use. You do need to understand assessment result sets to get valuable information out however or you will be reduced to finding out how many Italians eat sandwiches!

 

As it happens in will run on fairly ordinary kit but integrated graphics are of variable quality.

 

And a great tool I'm sure it will eventually be Phil once some of the teething problems are sorted, the problem with the performance issue really comes down to how many schools will be able to afford to make the necessary upgrades to run it though, that's where I believe it starts to become an issue for yourself, as if other MIS providers start offering similar tools (which they probably will now you are!) that don't require schools to fork out thousands for new hardware, that could potentially sway the competition...

 

[all of course my own opinion!]

[synaesthesia]

I don't think its abilities are being questioned - the input from our SIMS admins has been excellent so far and they've been chomping at the bit to get it. I think in a very vague nutshell, the question is this.

 

The same results could be achieved without anywhere near the amount of resources used. It is programmatically (woo is that even a word?) separate from SIMS so there's no need to rely on similar technology (after all, who'd enjoy an addon to FMS that's written in the same language? ) So, as pretty as it looks, why this rather than alternatives? Why not c++ if you want to compare to games directly, referring to an earlier post. Really pushing it a bit, but why not something along the lines of an Excel driven macro-fest? (ala Target Tracker).

Why not even Flas.... oh yeah, iGroans

[PhilNeal]

That's a great reply - thanks synaesthesia!

 

.net is pretty fast - we do have some "technical legacy" to deal with which will speed up things again in the autumn release but it isn't bad today and certainly a lot faster than it was when it was on beta.

[pcstru]

I suppose being on EduGeek you're interested in the technology. I'm interested in what something can do and I've seen enough revelations in schools to know that this is a very powerful tool.

 

The development team's brief was not to make it run on a hairdryer; it was to produce a very powerful tool that was really easy to use. You do need to understand assessment result sets to get valuable information out however or you will be reduced to finding out how many Italians eat sandwiches!

I'm interested in many things. The aim to provide a piece of software that is easy to drive and allows people in schools to easily analyse performance data, is a laudable aim. I think the decision to do that and the subsequent effort to realise it, should be applauded. From what little I have seen, it does look 'powerful' (flexible) and reasonably 'intuitive' (as these things go).

 

Having some experience of software development (20+ years), from lone coder to managing reasonably sized multi project development programmes, I'm not aware of ease of use ever having being a bar to system efficiency. More the reverse seems true, good design is sometimes negated by sloppy implementation making the user experience 'clunky' or leading to failure because the user experience is : "it's so advanced it doesn't even work yet!". On the other hand, first releases of products are often compromises and performance problems can be addressed in future versions; providing of course people recognise there is a problem and are able to correctly identify what the actual problem is - rather than (say) just expecting customers to throw more hardware at it.

 

As it happens in will run on fairly ordinary kit but integrated graphics are of variable quality.

 

Variable quality - enough to struggle to draw a pie chart?!

[synaesthesia]

Cheers Phil. I do remember seeing it at a Capita conference and it was a little slow then (running on a laptop)

Obviously with so many variations of machinery and OS it'll run on no doubt there's going to be some combinations that won't work quite as well as others. Hopefully in our case we can narrow down what's dragging it back a little and then nail it

[PiqueABoo]

Performance, especially how quick you can update the the screen with whatever, is a key factor in ease-of-use - always has been, always will be, but..

 

first releases of products are often compromises and performance problems can be addressed in future versions

 

..I think that is where it often all goes wrong. Manglers, sales-droids et al who just don't get behind the scenes that well respond to bells & whistles hitting their screens quickly (so they can make important exec decisions about colours, fonts and the like), and getting those there and kind-of-working all too often takes precedence over a really well-considered underlying architecture. It really doesn't take long before the "weight" of it all makes it increasingly difficult to go back and do what you should have taken your time over in the first place i.e. performance fixes are often compromised by limited 'wriggle-room" later down the line. I'll leave how that could possibly be true in the fabulous, brave new world of reusable, extendable, flexible, abstract OO-everything, as an exercise for the reader.

 

IME, which is similar, devs have long had that tendency to shrug and respond "throw more h/w at it" - heard it over and over forever, but it's by no means unique to Capita. That said, it's been much more common for server-side - the pragmatic realities on client-side have generally been better recognised and better respected.

 

.net is pretty fast

 

dotNet can be pretty fast, and generally sufficiently fast, serious number-crunching (which this scenario clearly isn't) excluded. I'm generally happy to write in pure dotNet for user-land code now, but were someone else of the current generation to implement the same code, I would expect it to go around the houses quite a bit e.g. why write a line to add two and two together, when you can implement an ever so clever, general purpose arithmetic engine, which of course despite the ever-lasting intention will still get thrown away within a few years to make way for the latest new and exciting dev-tech.

 

SQL has a lot to answer for re. resources and performance too, but if customers have an "easy/standard access to our data" tick-box there's not much choice.

[pcstru]

SQL has a lot to answer for re. resources and performance too, but if customers have an "easy/standard access to our data" tick-box there's not much choice.

 

I don't think that's the fault of SQL. My experience is that programmers are often skilled at procedural programming but treat SQL as if it is a just another procedural programming tool. They write some SQL and it gives the results they expect, they don't care about the cost and often enough in a development environment, aren't aware of the cost in a production environment. This can be exacerbated by naive developers using multiple transactions (ok, first we will ask the database for a list of students, then we will take that result set as an array and ask the database to give us the subject results for each of those students, one by one) where one transaction will do the job. Lack of database skills also leads to poor schema design and poor index selection. Well designed relational databases - even ones run on MS SQL, are quite capable of churning through results sets consisting of millions of rows in very short order - IF they are well designed and implemented. If not, go make a coffee or two.

[RobG]

I'd like to drag this back to the original question from synaesthesia about hardware.

 

Our data manager/exams officer went for training/demo on this and due to the fact that the laptop they demoed it on struggled, he's asked for a new pc for it.

 

His pc is getting on a bit, it's a 3ghz PentiumD (the original dual core pre-c2d, just a pair of p4's jamed onto the substrate under the IHS) 2gb of ram and agp x4 ATi Radeon 7200 (yes the original radeon which is DX7)

 

the great info given to him about minimum spec was

Core2Duo or AMD equivalent

4gb ram

A separate graphics card.

as a technical spec that is a total joke.

Going by that it should work, it's a dual core cpu, 2gb of ram is a bit low but should work and it has got a separate graphics card.

 

I called up the Captia "help desk" about this and they could tell me nothing more than that.

 

So we going to overspec, spoken to our main supplier and we looking at getting him a SB i5 2400, 8gb ddr3 1333mhz and . . .

I was thinking gts250 for cost, cores and raw performance it's great value for money, they've suggested a Quadro FX380 instead

It's a far weaker card (they are just 9500gt) however being a workstation card the drivers and support do make a major difference in most non-gaming applications (a quadro will perform around 2x-4x better than a gaming card with CAD or 3d rendering software)

 

Anyone know if there's any advantage to a workstation card over a consumer card with discover?

[localzuk]

I'd like to drag this back to the original question from synaesthesia about hardware.

 

Our data manager/exams officer went for training/demo on this and due to the fact that the laptop they demoed it on struggled, he's asked for a new pc for it.

 

His pc is getting on a bit, it's a 3ghz PentiumD (the original dual core pre-c2d, just a pair of p4's jamed onto the substrate under the IHS) 2gb of ram and agp x4 ATi Radeon 7200 (yes the original radeon which is DX7)

 

the great info given to him about minimum spec was

as a technical spec that is a total joke.

Going by that it should work, it's a dual core cpu, 2gb of ram is a bit low but should work and it has got a separate graphics card.

 

I called up the Captia "help desk" about this and they could tell me nothing more than that.

 

Sorry to break it to you, but when a company gives you a minimum spec of C2D or AMD equiv and 4GB RAM, a Pentium D and 2GB of RAM is not 'a bit low'! And expecting a DX7 era card to be fine for a program released this year is expecting a lot also.

 

That tech spec is perfectly fine from what I can see!

[BatchFile]

Not overly impressed with Discover so far to be honest. I thought the graphs were going to be much more flexible than they are. Anyone else feel the same?

 

Yup - falls on it's @r5e by not being able to do behaviour and achievement points between dates... (Ok I suppose I could do an excel macro report to count up the points between dates then import them into an AM aspect - but come on!)

[matt40k]

@BatchFiles - it's the first major release, give it a few years and it'll do more fancy things then you can shake a stick at.

[LosOjos]

I'd also like to suggest openly to any Capita employee's following this thread (@PhilNeal I mean you ) that you add "64-bit Windows Environment or Windows Server Environment" to the list of minimum requirements. The reason? You can't have 4GB of RAM and a dedicated graphics card without them, 32-bit Windows (with the exception of server versions) are limited to 4gb of RAM total, that includes system memory and graphics memory.

 

That being said, I now have Discover running on my dual core 3.2Ghz Intel with 4GB RAM (3GB accessible) and 1GB Geforce GT240 and it runs flawlessly

Edited July 15, 2011 by LosOjos
Pedantry

[CyberNerd]

@BatchFiles - it's the first major release, give it a few years and it'll do more fancy things then you can shake a stick at.

 

you think silvershite will be around in a few years? MS seem to be dropping it like a hot potato in favour of HTML5/javascript to keep up with Apple/Google. I'm sure Capita will happily pass on the redevelopment costs to schools for the next (low system requirement, platform independent, browser based) version.

[RobG]

Sorry to break it to you, but when a company gives you a minimum spec of C2D or AMD equiv and 4GB RAM, a Pentium D and 2GB of RAM is not 'a bit low'! And expecting a DX7 era card to be fine for a program released this year is expecting a lot also.

 

That tech spec is perfectly fine from what I can see!

It's a joke because c2d covers a rather large range of cpu's, go back and try using one of the old low end 1.8ghz c2d or one of the laptop 1.2ghz ones, there's a massive difference between those and a c2d running at 2.6ghz or higher

And it's nothing to do with "expecting" the card to work, I don't expect the card to work, but listing "A separate graphics card." (and yes that is the exact wording on the documentation I got for the technical spec.) they don't say dx9 minimum.

So yes is a TOTAL JOKE, is like listing the minimum requirement to enter a Formula 1 race as "a car"

 

I'll also point out that sme of the top end onboard/ondie gpu's are actually better than some bottom end graphics cards.

Edited July 15, 2011 by RobG

[TechMonkey]

I'd like to drag this back to the original question from synaesthesia about hardware.

 

Our data manager/exams officer went for training/demo on this and due to the fact that the laptop they demoed it on struggled, he's asked for a new pc for it.

 

His pc is getting on a bit, it's a 3ghz PentiumD (the original dual core pre-c2d, just a pair of p4's jamed onto the substrate under the IHS) 2gb of ram and agp x4 ATi Radeon 7200 (yes the original radeon which is DX7)

 

the great info given to him about minimum spec was

as a technical spec that is a total joke.

Going by that it should work, it's a dual core cpu, 2gb of ram is a bit low but should work and it has got a separate graphics card.

 

I called up the Captia "help desk" about this and they could tell me nothing more than that.

 

So we going to overspec, spoken to our main supplier and we looking at getting him a SB i5 2400, 8gb ddr3 1333mhz and . . .

I was thinking gts250 for cost, cores and raw performance it's great value for money, they've suggested a Quadro FX380 instead

It's a far weaker card (they are just 9500gt) however being a workstation card the drivers and support do make a major difference in most non-gaming applications (a quadro will perform around 2x-4x better than a gaming card with CAD or 3d rendering software)

 

Anyone know if there's any advantage to a workstation card over a consumer card with discover?

 

At a meeting I was at Wednesday someone was saying they had it running with an integrated graphics chip on an i.., 5 but could have been a 7. I'll be happier with that as a decent chip and lots of RAM is really a given today, a dedicated graphics card is odd in my mind.

 

The issue is they openly admit they don't know the min spec as they have had all sorts of results on differing hardware. So the only advice seems to be test it and see if it works.

[CAM]

So how essential is a graphics card? I'm stuck as I have been asked to find a PC near the £300 mark but they all come with on-board or insufficient RAM.

[BatchFile]

It'll run on most things that would be up to spec to regularly run SIMS - it's just a bit jerky flipping graphs over.

[RobG]

I'd also like to suggest openly to any Capita employee's following this thread (@PhilNeal I mean you ) that you add "64-bit Windows Environment or Windows Server Environment" to the list of minimum requirements. The reason? You can't have 4GB of RAM and a dedicated graphics card without them, 32-bit Windows (with the exception of server versions) are limited to 4gb of RAM total, that includes system memory and graphics memory.

 

That being said, I now have Discover running on my dual core 3.2Ghz Intel with 4GB RAM (3GB accessible) and 1GB Geforce GT240 and it runs flawlessly

Yes that's very true as well, you'll lose from the 4gb roughly 250mb (due to motherboard addressable resources) +video card ram, so with a graphics card with 1gb of vram you're generally down to around 2.75gb system ram with only a 32bit OS.

 

Nice to here the gt240 is doing well for you.

 

 

At a meeting I was at Wednesday someone was saying they had it running with an integrated graphics chip on an i.., 5 but could have been a 7. I'll be happier with that as a decent chip and lots of RAM is really a given today, a dedicated graphics card is odd in my mind.

 

The issue is they openly admit they don't know the min spec as they have had all sorts of results on differing hardware. So the only advice seems to be test it and see if it works.

That's sort of my exact point, not having had experience of this before and they themselves don't seem to know, so I came here to ask others, esp on the point of workstation vs consumer card.

[skunk]

Just as we are deploying these into our offices. Should be fun.

 

If it doesn't run then we shall wait for the govt. to increase school funding again before using Discover. Would be a shame, it is supposed to be nice.

[PhilNeal]

Here's a response from one of the experts!

 

"Its a tricky one to phrase as you almost always want to fit RAM in matched pairs which means that most 32-bit systems have 4Gb of RAM (2x2Gb or 4x1Gb) in them but do only have 3Gb accessible. Hence why RAM recommendations on software is usually 1Gb, 2Gb or 4Gb.

 

You can confirm that 3Gb is sufficient for Discover to run well (it is!) and you do not need a 64-bit operating system. I'll look at the phrasing of the minimum requirement"

[synaesthesia]

Linda from Capita showed the Northants techies it yesterday and this also came up. It ran OK on her laptop (which I know full well is a 4GB Core2Duo based machine but it does NOT have a discrete graphics chip) - however I was keen to stress it was running locally - there's not just the graphical interface to consider but the network infrastructure too - it has to pull the data from somewhere. It also needs another SQL instance which is slightly extra load on your SIMS server but it doesn't do a huge amount; unless your SIMS server is just about due for replacement anyway then you shouldn't have a problem.

 

Back to client-side though, you will EASILY find a machine that will run it at it's "best" for £300.

I think the term "discrete graphics" needs to be defined more carefully in their specifications. A Core i3 or i5 has a GPU builtin which is more powerful, as I've mentioned, then any card you'll buy for 30 or 40 notes, and is more than powerful enough to run Discover. Our new suite of Stone PC's costs just over 300 each with monitors.

Extra RAM is no big deal - you're WELL under a tenner a gig these days.

 

I will warn you this though, for expectation reasons.

Out of interest, I tested this on a quad core, 16GB ram system with a 1GB ATI 5770 graphics card. It ran slowly. I think we have certain expectations after playing Call of Duty flawlessly at 60FPS that things will be uber-smooth. It doesn't work like that, it's still early days and it's running in-browser.

 

2GB ram is more than enough.

"Core2 Duo" , as mentioned, is not limited to things with "Core2" badged on them. As time goes on, most the "Pentium D's" you see are in fact Core2 kit which has been rebadged and remarketed for the modern era.

 

Think the first time you've seen something dreadful and silverlight based like Eclipse.net or Junior Librarian.net. There is your benchmark.

 

 

*** Edit : I should also hasten to add, in Capita's defense (unusual for me!!) that people seem to be acting in recoiled horror at the specs. Discover is *NOT* a must-have. It is not a requirement for you, your school, your LEA or your children's parents. It is an extra that you may wish to use. I think people's opinions have been made clear and the same job could be done with less graphical giff-gaff. Would it be as easy to use? Would it be as intuitive? Would you be able to drag and drop anything onto anything else if you had the same thing in an Access database?

Edited July 16, 2011 by synaesthesia

[matt40k]

@PhilNeal are you quoting Ben?

 

I will warn you this though, for expectation reasons.

Out of interest, I tested this on a quad core, 16GB ram system with a 1GB ATI 5770 graphics card. It ran slowly. I think we have certain expectations after playing Call of Duty flawlessly at 60FPS that things will be uber-smooth. It doesn't work like that, it's still early days and it's running in-browser.

 

I think you'll find it because it doesn't want to be a business computer

[matt40k]

Best bit about Discover is that it uses the existing SIMS data, so you're not limited by what you can export from CTF, or a larger cost because they use Capita SIMS Business Objects\API.

[Yogi72]

we have been using and demonstrating discover on a 2yr old cheap acer laptop for the last 2 months..

 

I found the trick was to ensure sims is not open i.e use the discover login directly..takes a lot less resource..

 

On a seperate note...does anyone know what the parameters of the attendance timeline are i.e cumulative or weekly calculated.

 

On a final note..discover presented to 20 schools so far.. 95% take up rate..especially useful on identifying behavioural issues by category and location..i.e bullying.