Predictika · 2U two-socket Intel Xeon rack server configurations

An AI sales agent's 2U rack server build, checked before it goes out

An AI sales agent prepares a configuration for a 2U two-socket rack server: SKU, processors, memory, drives, PCIe cards and power supplies. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the configuration limits and returns a verdict, every reason and the figures it computed.

Scope: Acer Altos BrainSphere R389 F4 2U rack servers (12 x 3.5-inch, 12 x 3.5-inch GPU and 24 x 2.5-inch SKUs): processors, memory, drives, PCIe cards and power supplies. See everything this validator covers

This validator models configurations of the Altos BrainSphere R389 F4 (Altos Computing, an Acer subsidiary) 2U Intel Xeon Scalable rack server, not just the scenarios shown below.

  • SKUs and drives — 12 x 3.5-inch standard, 12 x 3.5-inch GPU (which adds 2 rear 2.5-inch bays) and 24 x 2.5-inch standard (4 U.2 + 20 SAS/SATA bays); the U.2 mezzanine card takes one of the 2 OCP mezzanine slots.
  • Processors and memory — two sockets, processors up to 205 W TDP (165 W on the GPU SKU), 12 DIMM slots per processor, RDIMM up to 64 GB and LRDIMM up to 128 GB per module; memory capacity and drive count are computed and compared with what the quote states.
  • PCIe cards and power — 6 full-height half-length slots (2 of them x16) and 2 low-height half-length slots (1 of them x16); a 1+1 redundant pair of 800 W or 1200 W Platinum power supplies.

Any R389 F4 configuration that touches these rules is in scope — not only the scenarios pictured here. Not covered: prices (no public price list, so this validator produces no price build-up or line-item quote), specific CPU, DIMM, drive, GPU and RAID card part numbers, slot lane sharing (x0 or x8 options), other Altos servers and workstations, and Acer consumer or commercial PCs.

The AI sales agentBuilds the server configuration from the conversation and says it is ready.
Predictika validatorChecks every rule at once and returns the verdict, each finding, and the computed figures.

7 scenarios in 3 groups · each group runs from the agent’s first attempt to the outcome · click one to switch

What the agent proposes

The AI sales agent says

Illustrative agent statement. The validator receives the structured request below.

Proposing agent: AI sales agent (scripted)

Before sending, the agent must call Recorded run

Predictika validator


      

        
        
        
      

      
Waiting for the call.

      
    

  

  
  

    
Where rules-based CPQ gets it wrong
Click to expand
    
Our assessment of what a typical rules-based CPQ (product rules, price rules and validation rules, as in Salesforce CPQ) does with the same cases. It is a judgment about effort and reliability, not a claim that nothing can be built.

    
What the validator caughtTypical rules-based CPQPredictika
Bays and slots that depend on the SKUOnly with care
A set of product rules per SKU. It works, but each SKU needs its own limits (front, U.2 and rear bays, processor power) kept up to date.		The SKU and every count are in one request, so each limit is checked against the SKU selected.
A mezzanine card that uses up an OCP slotOnly with workarounds
Using U.2 drives takes one of the two OCP slots, so the OCP card count depends on a drive choice on another line; this needs a cross-line calculation.		Slots needed are computed from the OCP cards and the U.2 drives on each call and compared with the 2 available.
A memory module limit that depends on the memory typeOnly with care
Modules of 64 GB are fine for RDIMM but not for 128 GB; each memory type needs its own size limit and the rule must read the type chosen elsewhere.		Memory type and module size are in one request, and the limit for the chosen type is applied on every call.
Stated memory and drive counts that must match the quoteOnly with care
A stated figure is a free-entry field; a rule can compare it with a computed total, but someone has to write and maintain that comparison.		The stated figures travel with the configuration and are compared with the computed figures on every call.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as the configuration is saved or a step is completed, so a rep often fixes one error and meets the next.		All findings come back together, whatever order the choices arrived in.

  

  

    
What this validator catches: findings from its test cases
Click to expand
    
Every rule is checked on every call, in any order the choices arrive; one problem never hides another. Blocking findings stop the quote; approval findings need a person to sign off; everything else is computed.

    
1The R389 F4 is a two-socket server; the quote specifies 3 as the processor count.
2This SKU supports processors up to 205 W TDP; the quote has a 250 W processor.
3This SKU supports processors up to 165 W TDP; the quote has a 205 W processor.
4The quote has 24 memory modules, but the processors on it provide only 12 DIMM slots (12 per processor).
5This memory type is supported up to 64 GB per module; the quote has 128 GB modules.
6This memory type is supported up to 128 GB per module; the quote has 256 GB modules.
7The quote has 21 front SAS/SATA drives, but this SKU has 20 front SAS/SATA bays.
8The quote has 13 front SAS/SATA drives, but this SKU has 12 front SAS/SATA bays.
9The quote has 4 U.2 drives, but this SKU has 0 U.2 bays (only the 24 x 2.5-inch SKU has 4).
10The quote has 2 rear drives, but this SKU has 0 rear bays (only the GPU SKU has 2).
11The quote has 3 rear drives, but this SKU has 2 rear bays (only the GPU SKU has 2).
12The 24 x 2.5-inch SKU takes 2.5-inch drives only, and 3.5-inch drives must be among the front drives; the quote lists 4 3.5-inch drives.
13The server has 2 OCP mezzanine slots and the U.2 mezzanine card takes one; the quote needs 3.
14The server has 2 full-height x16 slots; the quote has 3 full-height x16 cards.
15The server has 6 full-height half-length PCIe slots; the quote has 7 full-height cards.
16The server has 2 low-height half-length PCIe slots; the quote has 3 low-height cards.
17The server uses a 1+1 redundant power supply pair; the power supply count on the quote is 1.
18The power supplies are 80 PLUS Platinum 800 W or 1,200 W units; the quote has 550 W units.
19The quote states 512 GB of memory, but the modules on it add up to 768 GB.
20The quote states 26 drives, but the drives on it add up to 24.

  

  

    
Scenarios and results
Click to expand
    
The recorded result of the validator for each scenario on this page: the request the AI agent sent, the verdict, and every finding, in the order they appear above.

    
1. GPU server: A GPU server with 205 W processors

    
The agent's request. Please quote an Altos BrainSphere R389 F4, 12 x 3.5-inch GPU SKU, with 2 Intel Xeon processors of 205 W, 24 x 128 GB LRDIMM memory, 12 front drives, 2 rear drives, 1 OCP mezzanine card, 2 full-height x16 cards, 1 full-height x8 card, 2 x 1200 W power supply units. The quote states 3,072 GB of memory and 14 drives in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Processor power too high for this SKU. This SKU supports processors up to 165 W TDP; the quote has a 205 W processor.
    • 
    

    
2. GPU server: The agent resubmits after fixing the errors

    
The agent's request. Please quote an Altos BrainSphere R389 F4, 12 x 3.5-inch GPU SKU, with 2 Intel Xeon processors of 165 W, 24 x 128 GB LRDIMM memory, 12 front drives, 2 rear drives, 1 OCP mezzanine card, 2 full-height x16 cards, 1 full-height x8 card, 2 x 1200 W power supply units. The quote states 3,072 GB of memory and 14 drives in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Configuration summary. 3,072 GB of memory (24 x 128 GB); 14 drives; OCP mezzanine slots used: 1 of 2.
    • 
    

    
3. 24-bay node with U.2 drives: A 24-bay node with 3.5-inch drives and two OCP cards

    
The agent's request. Please quote an Altos BrainSphere R389 F4, 24 x 2.5-inch standard SKU, with 2 Intel Xeon processors of 205 W, 24 x 32 GB RDIMM memory, 20 front drives, 4 of them 3.5-inch, 4 U.2 drives, 2 OCP mezzanine cards, 2 full-height x16 cards, 4 full-height x8 cards, 1 low-height x16 card, 1 low-height x8 card, 2 x 1200 W power supply units. The quote states 768 GB of memory and 24 drives in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • 3.5-inch drives do not fit. The 24 x 2.5-inch SKU takes 2.5-inch drives only, and 3.5-inch drives must be among the front drives; the quote lists 4 3.5-inch drives.
    • 
      
  • Not enough OCP mezzanine slots. The server has 2 OCP mezzanine slots and the U.2 mezzanine card takes one; the quote needs 3.
    • 
    

    
4. 24-bay node with U.2 drives: The agent resubmits after fixing the errors

    
The agent's request. Please quote an Altos BrainSphere R389 F4, 24 x 2.5-inch standard SKU, with 2 Intel Xeon processors of 205 W, 24 x 32 GB RDIMM memory, 20 front drives, 4 of them 3.5-inch, 4 U.2 drives, 1 OCP mezzanine card, 2 full-height x16 cards, 4 full-height x8 cards, 1 low-height x16 card, 1 low-height x8 card, 2 x 1200 W power supply units. The quote states 768 GB of memory and 24 drives in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • 3.5-inch drives do not fit. The 24 x 2.5-inch SKU takes 2.5-inch drives only, and 3.5-inch drives must be among the front drives; the quote lists 4 3.5-inch drives.
    • 
    

    
5. 24-bay node with U.2 drives: The agent resubmits after fixing the errors

    
The agent's request. Please quote an Altos BrainSphere R389 F4, 24 x 2.5-inch standard SKU, with 2 Intel Xeon processors of 205 W, 24 x 32 GB RDIMM memory, 20 front drives, 4 U.2 drives, 1 OCP mezzanine card, 2 full-height x16 cards, 4 full-height x8 cards, 1 low-height x16 card, 1 low-height x8 card, 2 x 1200 W power supply units. The quote states 768 GB of memory and 24 drives in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Configuration summary. 768 GB of memory (24 x 32 GB); 24 drives; OCP mezzanine slots used: 2 of 2.
    • 
    

    
6. Large memory: A GPU server with 24 x 256 GB LRDIMM modules

    
The agent's request. Please quote an Altos BrainSphere R389 F4, 12 x 3.5-inch GPU SKU, with 2 Intel Xeon processors of 165 W, 24 x 256 GB LRDIMM memory, 12 front drives, 2 rear drives, 1 OCP mezzanine card, 2 full-height x16 cards, 2 x 1200 W power supply units. The quote states 6,144 GB of memory and 14 drives in total. It is ready to send.

    
Verdict: Blocked.

    
    
      
  • Memory module too large. This memory type is supported up to 128 GB per module; the quote has 256 GB modules.
    • 
    

    
7. Large memory: The agent resubmits after fixing the errors

    
The agent's request. Please quote an Altos BrainSphere R389 F4, 12 x 3.5-inch GPU SKU, with 2 Intel Xeon processors of 165 W, 24 x 128 GB LRDIMM memory, 12 front drives, 2 rear drives, 1 OCP mezzanine card, 2 full-height x16 cards, 2 x 1200 W power supply units. The quote states 3,072 GB of memory and 14 drives in total. It is ready to send.

    
Verdict: Allowed.

    
    
      
  • Configuration summary. 3,072 GB of memory (24 x 128 GB); 14 drives; OCP mezzanine slots used: 1 of 2.
    • 
    

  

  

    
Narration transcript
Click to expand
    
The text of the narrated walkthrough, step by step.

    
    
      
  1. Welcome to a demo of Predictika's Validator. An AI sales agent prepares a configuration for a 2U two-socket rack server: SKU, processors, memory, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of the configuration limits and returns a verdict, every reason and the figures it computed.
    2. 
      
  2. I will start with the first request by the agent.
    3. 
      
  3. The agent says. Please quote an Altos BrainSphere R389 F4, 24 x 2.5-inch standard SKU, with 2 Intel Xeon processors of 205 W. The quote states 768 GB of memory and 24 drives in total. The rest of the request is on screen.
    4. 
      
  4. Before it goes out, the agent calls the validator.
    5. 
      
  5. Blocked, with two findings, all in one answer. The 24 x 2.5-inch SKU takes 2.5-inch drives only, and 3.5-inch drives must be among the front drives; the quote lists 4 3.5-inch drives. The server has 2 OCP mezzanine slots and the U.2 mezzanine card takes one; the quote needs 3.
    6. 
      
  6. The validator returned two blocking findings in one answer. A check that reads one field at a time would pass most of this; it shows only when the choices in the quote are checked against each other.
    7. 
      
  7. What this shows: for a mezzanine card that uses up an OCP slot, a rules-based CPQ can do this only with workarounds.
    8. 
      
  8. Now the agent resubmits.
    9. 
      
  9. The corrected request. It changes the OCP mezzanine cards to 1.
    10. 
      
  10. It calls the validator again.
    11. 
      
  11. Still blocked, with one finding, all in one answer. The 24 x 2.5-inch SKU takes 2.5-inch drives only, and 3.5-inch drives must be among the front drives; the quote lists 4 3.5-inch drives.
    12. 
      
  12. One finding cleared. Still blocking: 3.5-inch drives do not fit.
    13. 
      
  13. The agent resubmits once more.
    14. 
      
  14. The revised request. It removes the of which 3.5-inch.
    15. 
      
  15. It calls the validator once more.
    16. 
      
  16. This time the validator clears it.
    17. 
      
  17. The earlier finding is cleared, and the validator's figure matches the quote.
    18. 
      
  18. Two problems caught in one call, fixed in two passes, and a clean quote. That is the value of Predictika's validator: catching mistakes before they reach the customer. Now you can try the other scenarios yourself.
    19.